The Brazilian state and transparency: evaluating compliance with freedom of information

The Brazilian State and Transparency

EvaluatingCompliancewith Freedomof Information

Gregory MichenerLuiz Fernando Marrey Moncau

Rafael Velasco

The Brazilian State and TransparencyEvaluating Compliance with

Freedom of Information

Edition produced by FGV’s Rio de Janeiro Law School

Praia de Botafogo, 190 | 13th floor

Rio de Janeiro | RJ | Brazil | ZIP code: 22250-900

55 (21) 3799-5445

www.fgv.br/direitorio

Gregory MichenerLuiz Fernando Marrey Moncau

Rafael Velasco

The Brazilian State and TransparencyEvaluating Compliance with

Freedom of Information

Catalographic card prepared by Mario Henrique Simonsen Library / FGV

Michener, GregoryThe Brazilian state and transparency: evaluating compliance with freedom of information / Gre-

gory Michener, Luiz Fernando Marrey Moncau and Rafael Velasco. – Rio de Janeiro : Escola de Direito do Rio de Janeiro da Fundação Getulio Vargas, Centro de Tecnologia e Sociedade, 2016.

130 p.

Em parceria com a EBAPE.Inclui bibliografia.

ISBN: 978-85-63265-68-5

1. Transparência na administração pública – Brasil. 2. Direito à informação – Brasil. 3. Brasil. [Lei n. 12.527, de 18 de novembro de 2011]. 4. Poder judiciário – Brasil. I. Moncau, Luiz Fernando. II. Velasco, Rafael Braem. III. Escola de Direito do Rio de Janeiro da Fundação Getulio Vargas. Centro de Tecnologia e Sociedade. IV. Escola Brasileira de Administração Pública e de Empresas. V.Título.

CDD – 341.3

FGV’s Rio de Janeiro Law School Edition

This Work is licensed under the Creative Commons license Attribution-ShareAlike 4.0 International

To view the full legal license visit:https://creativecommons.org/licenses/by-sa/4.0/legalcode

You have the right to:

Share — to copy and redistribute the material in any medium or format.

Adapt — remix, transform, and build upon the material for any purpose, including commercial.

The licensor cannot revoke these freedoms as long as you follow the license terms. Under the

following terms:

Attribution — you must give appropriate credit, provide a link to the license, and indicate if

changes were made. You may do so in any reasonable manner, but must not in any way suggest

endorsement from the licensor.

ShareAlike — if you remix, transform, or build upon the material, you must distribute your

contributions under the same license as the original.

No additional restrictions — you may not apply legal terms or technological measures which

would legally restrict others from use permitted by the license.

Printed in BrazilThis first edition was completed in June 2016This book is in the Legal Deposit Division of the Brazilian National Library.

Authors are responsible for all concepts and information presented in this book.

Coordination Rodrigo Vianna, Sérgio França, Thaís MesquitaCover Antonio ValérioDiagramation Antonio Valério Reviser Ana Silvia Mineiro

Table of Contents Acknowledgments 11

Executive summary 13

Introduction 19

Chapter 1 – Methodology 25

1. General Audit 25

1.1. Setting up the field experiment; user identities to test for bias 27

2. The Judicial Branch Audit 28

3. Preparations; procedural rules to ensure realistic responses 29

4. Metrics 30

4.1. Defining what constitutes a response 30

4.2. Evaluating response accuracy rates 31

4.2.1. Accuracy evaluation criteria 31

4.2.2. Double coding of responses 32

4.2.3. Testing the reliability of double coding 33

4.3. Timeliness of responses 34

Chapter 2 – Results 35

1. Introduction 35

2. Aggregate analysis 35

2.1. Jurisdictional level analysis 36

2.2. Analysis by branch of government 41

3. The field experiment; user identities and question groups 46

3.1. Results; descriptive statistics 48

3.2. Statistical tests 49

4. Conclusions 51

5. Appeals; their place in the study 52

5.1. Introduction; background and procedure 52

5.2. Successful appeals; overturning unsatisfactory responses 53

Chapter 3 – Best practices and barriers to transparency 55

1. Introduction 55

2. Best practices 55

3. Barriers to Public Transparency 56

3.1. Requirements that the request be submitted or retrieved in person 57

3.2. Rejection of the request based on a broad interpretation of the “undue additional work” clause 58

3.3. Transmitting information in formats not processable by computers 60

3.4. Aggregation of information in a pattern different from the original request 62

3.5. Implementation of potentially intimidating procedures 62

Chapter 4 – Audit of the platforms used to manage FOI requests 65

1. Introduction 65

2. Information access barriers: a lack of compliant digital platforms 65

2.1. Impossibility of filing an appeal 66

2.2. Character limits 66

3. Evaluation of digital platforms 66

3.1. Methodology of evaluation; what constitutes a good FOI digital platform? 67

3.2. Submission of a confirmation email 67

3.3. Submission of a notification of response 67

3.4. Availability of a specific section for filing appeals 67

3.5. Use of log-in procedure instead of confirmation/protocol numbers 68

3.6. Summary 68

4. Comparative analysis of digital platforms 68

5. Conclusions 70

Part 2 – Judicial Branch Audit 71

Chapter 5 – Results 73

1. Methodology 73

1.1. Audit coverage 73

1.2. Development and submission of requests 74

1.3. Topics of requests 74

1.4. User identities: simulating the request process experienced by a regular citizen 82

1.5. Submission of requests 82

1.6. Appeals 82

1. 7. Field experiment; testing responses to justified vs. non-justified requests 83

2. Results 84

2.1. Results by request category 84

Request Category 1 85







2.2. Detailed results 98

2.3. Analysis of the courts 101

Chapter 6 – Barriers to transparency and access 103

1. Bureaucratic barriers 104

2. Legal and regulatory barriers, and misuse of exemptions 106

2.1. Excessively broad interpretation of exemptions 106

2.1.1. Disproportionate or unreasonable requests 107

2.1.2. Generic requests or requests requiring additional work 108

2.2. Misinterpretation of exceptions; active transparency and the format of documents 109

2.2.1. Requests for information already available in active transparency platforms 110

2.2.2. Questions regarding data format 112

Chapter 7 – Evaluating the platforms used to manage freedom of information requests 117

Conclusion 119

References 127

Index of tables Table 1 – Subjects of requests and number of submissions for each request 26

Table 2 – Kappa scores scale 33

Table 3 – Kappa score of the General Audit 34

Table 4 – Kappa score of the Judicial Branch Audit 34

Table 5 – Distribution of requests by user identities 47

Table 6 – Distribution of requests between institutional and non-institutional identity users 47

Table 7 – Chi-square test of differences in response rate 49

Table 8 – Chi-square test of differences in accuracy rate 50

Table 9 – T-test of differences in average response time 51

Table 10 – Results of hypothesis testing 51

Table 11 – Platform evaluation 69

Table 12 – Distribution of audited courts 73

Table 13 – Request topics of the Judicial Branch Audity 75

Table 14 – Requests by identity 82

Table 15 – Courts by group 84

Table 16 – Court ranking 102

Index of figures Figure 1 – Aggregate Results 36

Figure 2 – Aggregate analysis of average response time 36

Figure 3 – Federal level results 37

Figure 4 – Average response time at the federal level 37

Figure 5 – State level results 38

Figure 6 – Average response time at the state level 39

Figure 7 – Municipal level results 40

Figure 8 – Average response time at the municipal level 41

Figure 9 – Results by branch of government 42

Figure 10 – Average response time by branch of government 42

Figure 11 – Analysis of executive branches by jurisdictional level: states and federal government 43

Figure 12 – Average response time of executive branches by jurisdictional level: states and federal government 44

Figure 13 – Analysis of executive branches by jurisdictional level: municipalities and federal government 45

Figure 14 – Average response time of executive branches by jurisdictional level: municipalities and federal government 46

Figure 15 – Field experiment results 48

Figure 16 – Average response time for field experiment 49

Figure 17 – Aggregate analysis results of Judicial Branch Audit 85

Figure 18 – Results of Request Category 1 86

Figure 19 – Average response time for Request Category 1 87













Figure 32 – Field experiment results: response and accuracy rates 99

Figure 33 – Field experiment results: average response time 100

A public transparency audit is an intensive undertaking and demands extreme

organization and coordination – especially in the case of two evaluations involving

dozens of jurisdictions, approximately 60 questions and more than 700 FOI re-

quests. Many people have contributed to this initiative since it began to take shape

over a year ago. Firstly, we would like to acknowledge the essential support provi-

ded by the Open Society Foundations (OSF) without which this work would pro-

bably not exist. The editors would like to thank the whole team who worked on this

project. Gratitude goes to Karina Rodrigues Furtado from the Brazilian School of

Public and Business Administration (EBAPE) who, as research assistant to Gregory

Michener, examined the methodology and results of each evaluation of transparen-

cy in Latin America since 2004 – a task which ultimately inspired the Transparency

Audit Network and the partnership between the Getulio Vargas Foundation (FGV)

at EBAPE and the Center for Technology and Society at the FGV Law School, Rio

de Janeiro (FGV DIREITO RIO). Many people were involved in helping to develop,

gather and provide feedback on audit issues, including Joaquim Falcão, Marilia Fer-

reira Maciel, Pedro Mizukami, Karina Rodrigues, Marina Barros, Irene Niskier, Pedro

Acknowledgments

12The Brazilian STaTe and TranSparency

evaluaTing compliance wiTh Freedom oF inFormaTion

Augusto Pereira Francisco, and Vitor Coff del Rey. We thank Marina Barros, Karina

Rodrigues, and Jamila Venturini for writing parts of the report and for contribu-

ting to the data analysis. Additionally, we thank the students of the FGV DIREITO

RIO Jonas Coelho, Luiz Carlos Penner Rodrigues, and Pedro Henrique Lourenço da

Costa for their assistance in submitting Freedom of Information (FOI) requests and

appeals throughout the Judicial Branch Audit. We are also grateful to the people

who read early drafts of this report, including Pedro Mizukami, Ivar Hartmann, Pablo

Cerdeira, Diego Werneck, Fernando Leal, Maria Candida Rolim, and Luiz Antonio

Jucá Serão; as well as those who helped with coding, calculations, proof reading,

and organizing data: Tatiana Carli Mota, Marcelo Desterro Queiroz, Rebecca Davies,

and again, Luiz Antonio Jucá Serão. We also thank Ernesto Vilchis for the English

translation, Felipe Labouriau, who was responsible for the report’s graphics, Laura

Lydia, responsible for the cover’s art work, and Rachel Bevilacqua Daza, and Livya

Lemos Bandeira de Mello, who helped with the seminar’s logistical and design re-

lated work. Finally, we thank both the Vice Director of EBAPE/FGV, Álvaro Cyrino,

and the Director of FGV DIREITO RIO, Joaquim Falcão, for their unswerving logis-

tical and moral support.

Gregory MichenerAssociate Professor, EBAPE1/FGV2

Director, Program for Public Transparency (PTP)3, FGV

Luiz Fernando Marrey MoncauManager

Center for Technology and Society (Centro de Tecnolo-

gia e Sociedade)4, FGV DIREITO RIO5

Rafael VelascoProgram Coordinator

Program for Public Transparency, FGV

1 Brazilian School of Public and Business Administration or Escola Brasileira de Administração Pública e de Empresas.

2 Getulio Vargas Foundation or Fundação Getulio Vargas.

3 Public Transparency Program.

4 Center for Technology and Society.

5 Getulio Vargas Foundation Law School, Rio de Janeiro.

In 2011, Brazil became the 13th country in Latin America and 91st in the world to

grant its citizens specific mechanisms to guarantee the right to public information. Law

12.527, Brazil’s new Freedom of Information (FOI) law went into effect in May 20121.

It regulates several provisions in the federal Constitution, including section XXXIII of

Article 5 and Article 37 section II §3. The FOI law applies to all branches and levels of

government, including state owned companies and banks, among other entities.

Research Questions

Are Brazilian public institutions complying with the Freedom of Information Law?

What does the level of Compliance tell us about the need for greater commit-

ments, reforms and better practices?

Freedom of information is a fundamental right and recognized as ‘the oxy-

gen of democracy’2. Likewise, government transparency helps professionalize the

public administration and consolidate a rule of law based on accountability and

democratic openness.

Despite the importance of transparency, compliance with transparency statutes

is extremely onerous in political terms. Within this context, this study assesses the

degree of compliance with freedom of information in Brazil, examining commitments

to passive transparency – responses to requests for information. Additionally, the re-

search evaluates the procedures that citizens must follow in requesting information.

For an entire year, the Public Transparency Program (PTP) at the Getulio Var-

gas Foundation’s Brazilian School of Public and Business Administration (EBAPE)3

1 In Portuguese the law is known as ‘Lei de Acesso à Informação’ or LAI.

2 Phrase from the non-governmental organization Article 19.

3 Programa de Transparência Pública, Escola Brasileira de Administração Pública e de Empresas.

executive summAry



as well as the Center for Technology and Society (CTS)4 at the Law School of Rio de

Janeiro (FGV DIREITO RIO)5, conducted two transparency audits using a simulated

user methodology. More than 700 FOI requests were sent to almost 170 public enti-

ties from all three government branches and all levels of government; federal, state,

and municipal.

The first objective of the audit was to verify whether public entities are respon-

ding to FOI requests, and whether responses (i) comply with stipulated timeframes

and (ii) accurately respond to requests. The study also sought to understand whe-

ther public entities are complying with general norms of non-discrimination when

providing access to information. In other words, whether public entities’ responses to

information requests varied based on who submitted, and how the request was pla-

ced. Finally, the audit aimed to assess the existence and quality of digital platforms

provided so that citizens might access public information in a simple manner, using a

single platform to make requests and appeals, as well as to receive responses.

Report Outline

This report is divided into two parts. The first presents what is referred to as

‘the General Audit’, in which more than 450 FOI requests were submitted to the three

branches of government in eight jurisdictions at the federal, state, and municipal

levels, as well as state enterprises and constitutionally autonomous institutions. The

second part of the report is referred to as ‘the Judicial Branch Audit’, which consists

of a specific analysis of the Brazilian Judicial Branch. This audit involved over 260 FOI

requests sent out to more than 40 tribunals.

Methodology

The General Audit as well as the Judicial Branch Audit used the same metho-

dology, which was developed with the goal of being easily understood and replica-

ted by other institutions or persons interested in analyzing the level of government

compliance with FOI laws.

The methodology was developed in collaboration with numerous professors

and researchers from FGV’s EBAPE and FGV DIREITO RIO, and was based on an ex-

4 Centro de Tecnologia e Sociedade.

5 FGV – Fundação Getulio Vargas.

15EXECUTIVE SUMMARY

tensive review and analysis of audits in other countries. As will be shown in Chapter

1, our methodology used three straightforward metrics: response rate, accuracy rate,

and average response time. The response rate consisted of the proportion of re-

quests responded to, by a particular entity. The accuracy rate measured the propor-

tion of requests that were answered with relevant responses. The average response

rate calculated the number of days between the date the request was submitted and

the date of response by the various entities. All these metrics are indicators of the

institutional commitment to compliance with freedom of information norms.

Part I – General Audit

A total of 453 FOI requests were submitted based on 55 different questions,

targeted at 133 public entities across the three branches of government in eight

jurisdictions. The jurisdictions included the federal government, the Federal District,

and the states of Minas Gerais, Rio de Janeiro and São Paulo, as well as their respec-

tive capitals: Belo Horizonte, Rio de Janeiro and São Paulo.

The overall objective of the audit was to provide an overview of the practical

workings of Law 12.527/11. Three main thematic topic areas, all relevant to Brazil’s

current public policy challenges, were selected as the basis for requests: (i) gover-

nance and watchdog institutions; (ii) public participation and human rights; and (iii)

socioeconomic rights. The General Audit was exploratory in character, employing

a wide diversity of requests as a means of gauging the general responsiveness of

governmental entities, the application of the law and, consequently, governmental

commitments to transparency.

Of 453 requests, some 322 were submitted by four volunteers, who formed

two groups or user identities, each one comprised of a male and a female. Requests

were divided into two groups, one of institutional user identities; and the other of

non-institutional user identities. The former was composed of researchers from the

FGV – a fact that could be easily verified through a search of the person’s name on

the Internet. The volunteers with non-institutional identities had no trace of their

names on the Internet, and could therefore not be associated with any institution.

Based on discussions with public administrators responsible for FOI requests, FGV

researchers inferred that governmental authorities often use Internet searches of

requesters to identify details about the person making the request. This practice



creates an opportunity to consciously or unconsciously discriminate in the provision

of a public service. It was our objective to assess the extent of discrimination throu-

gh a field experiment.

As will be explained in Chapter 2, the findings of the General Audit show a

moderate to low rate of compliance with Law 12.527/11, however the results vary

widely across the various jurisdictions analyzed. The overall response rate was 69%,

the overall accuracy rate, 57%, and the average response time was 21 days. Despite

being slightly higher than the response period of 20 days established by the FOI law,

the average response rate is within the statutory maximum limit of 30 days applica-

ble to cases with justifiable time extensions.

In summary, the data suggests that the audited jurisdictions do not respond to

one out of three FOI requests received, and slightly more than half of the requests

receive responses that meet the minimum accuracy threshold.

It is important to note that we find wide variations in the degree of compliance

with the law across eight jurisdictions in the study. The State of Rio de Janeiro and

the Municipality of Rio de Janeiro exhibit an alarmingly low rate of compliance, with

response rates of 27% and 38% respectively. The situation is worse when we take into

account the accuracy rates of 17% for the state and 18% for the municipality. These

results, well below the study average, are partly due to decrees which illegally regulate

federal Law 12.527/11. For example, Decree 43.597 of 2012 in the state of Rio de Janeiro

requires that citizens fill out requests in person at the relevant agency, and sign a liabili-

ty statement. These requirements egregiously deviate from international norms as well

as Article 11 of the federal law; Law 12.5271/11 stipulates that citizens can request infor-

mation “by any legitimate means”. On the other hand, the response rates of the federal

branches of government, the Federal District, and the State and Municipality of São

Paulo, were quite high, at approximately 80%. Accuracy rates in these jurisdictions ran-

ged between 62%, in the case of the Municipality of São Paulo, to 76% for the federal

government. These findings show significant variation in commitment to the FOI law.

The results of the user identity experiment point to practices of discrimination

in responding to FOI requests. Specifically, the non-institutional users obtained a

response rate 10% below the response rate for the users with institutional identities

(64% vs. 74%), a remarkable difference that is statistically significant at the 10% level.

The non-institutional users also experienced much longer response times, an average

of eight days longer than institutional users (25.5 days vs. 17.5 days). Particularly alar-

17EXECUTIVE SUMMARY

ming, the female non-institutional identity had a very low response rate (57%), and

an accuracy rate of only 45%. Taking these figures into account, we believe the requi-

rement in Law 12.527/11, that citizens provide their real names and an identity number

(RG or CPF) should be immediately amended6. The use of real identities can result

not only in discrimination, but also intimidation. In short, only contact details should

be required to request information. The range of topics addressed in the FOI requests

submitted in the General Audit, as well as the number of entities analyzed, was consi-

derably broad. Nevertheless, the results allow us to make important inferences about

compliance with the FOI law in jurisdictions known to set examples across Brazil.

Part II – Judicial Branch Audit

The objectives of the Judicial Branch Audit were: (i) to evaluate compliance

with the FOI law by the Brazilian courts, (ii) to add new dimensions to the recent

discussion of this topic7, and (iii) to identify opportunities and challenges for more

efficient public management. In contrast to the General Audit, the Judicial Branch

Audit analyzed a limited number of topics in greater detail. These topics reflected

some of the current challenges faced by the Judicial Branch in Brazil, such as sa-

laries, the promotion of judges, the prevention of nepotism, and the administrative

management of the courts.

Seven separate FOI requests concerning these topics were submitted to 40 di-

fferent courts, comprising 264 requests in total. We assessed three Superior Courts

(Tribunais Superiores), including the Supreme Federal Court (Supremo Tribunal Fe-

deral, or STF)8, the Superior Court of Justice (Superior Tribunal de Justiça, or STJ)9,

6 The General Registry or Registro Geral (RG) is the official identity document in Brazil. The Natural Persons Register or Cadastro de Pessoas Físicas (CPF) is a number attributed by the Brazilian Federal Revenue Service to both Brazilians and resident aliens and is equivalent to a social security number.

7 See: ARTIGO 19, Balanço de 1 Ano da Lei de Acesso à Informação Pública: acesso à informação e os ór-gãos de justiça brasileiros – 2012-2013. São Paulo: Artigo 19, 2013. Available at: http://artigo19.org/wp-content/uploads/2014/01/balanco_1_ano_ lei_acesso.pdf [Accessed on: 03.11.2014], as well as, BEGHIN, N. & ZIGONI, C. (orgs), “Avaliando os websites de transparência orçamentária nacionais e sub-nacionais e medindo impactos de dados abertos sobre direitos humanos no Brasil”. Brasília: Instituto de Estudos Socioeconômicos, 2014. Availa-ble at: http://www.inesc.org.br/biblioteca/publicacoes/textos/pesquisa-transparencia-orcamentaria-nos-web-sites-nacionais-e-sub-nacionais [Accessed on: 29.10.2014]

8 Supreme Federal Court or “Supremo Tribunal Federal”.

9 Superior Court of Justice or “Superior Tribunal de Justiça”.



and the Superior Labor Court (Tribunal Superior do Trabalho, or TST)10, as well as

five Federal Regional Courts (Tribunais Regionais Federais, or TRF), five Regional

Labor Courts (Tribunais Regionais do Trabalho), 26 State Courts (Tribunais de Justi-

ça dos Estados), in addition to the Court of Justice of the Federal District and Terri-

tories (Tribunal de Justiça do Distrito Federal e Territórios).

The aggregate results show that out of a total of 264 FOI requests, 160 (61%)

received a response, of which only 69 (26%) were deemed accurate according to

the definition established in the methodology. The response rate did not show sig-

nificant variability across the seven requests submitted, hovering around 65%. The

accuracy rate of the responses was low relative to the rate observed in the General

Audit. However, the accuracy rate varied significantly across the courts analyzed,

between 5% and 62%, depending on the question submitted. When ranked based

on their performance in supplying the specific information requested, the thirteen

courts which obtained best accuracy rates were: the TST, SFT, TRF 4TH Region,

TRT 15th Region, TRT 1st Region, TJ-RR, TRT 10th Region, TRF 2nd Region, TRF 3rd

Region, TRT 3rd Region, STJ, TJ-RJ and TJ-DFT. It is worth highlighting that only

three state judicial courts rank among the courts with the best accuracy rates. Re-

garding the timeliness of their response, only four of the 40 courts did not comply

with the 20-day deadline stipulated by the law. The average response time was 14

days; below the average response time observed for those entities assessed in the

General Audit. The Freedom of Information Law establishes that a denial must occur

only under extraordinary conditions defined by law. The findings of the qualitative

assessment, however, indicate that FOI requests were repeatedly rejected, based on

excessively broad, and at times illegitimate interpretations of the permitted legal

exceptions. In other situations, the courts hampered access by creating bureaucratic

obstacles, which is forbidden by Law 12.527/11.

Furthermore, the vast majority of the courts analyzed did not have specific

platforms for the submission of FOI requests. Instead, they were found to employ

the websites of their ombudsmen.

Considering all of the above, we recommend that the efforts of the courts and

the National Judicial Council (CNJ) focus on a few key implementation objectives,

detailed at the end of the next section.

10 Superior Labor Court or “Tribunal Superior do Trabalho”.

Context

In recent years a wave of Freedom of Information (FOI) laws has swept across

the world, with more than two thirds of the approximately one hundred FOI laws

having been adopted in the last decade. In Latin America, only Venezuela and Cos-

ta Rica do not have some type of FOI legislation: Brazil was one of the last to join

this move towards greater transparency, approving its Freedom of Information Law

12.527 in 20111. This law delineates and makes effective the right guaranteed by Ar-

ticle 5, XXXIII, and section II of § 3rd of Article 37, of the federal Constitution, which

grants any citizen the right to obtain government information.

The FOI law complies with the decisions of the Inter-American Court of Human

Rights, which affirm the right of access to public information (Chile v Claude Reyes et

al2 and Gomes-Lund v Brasil)3. It also represents compliance with other treaties and

conventions, such as Article 19 of the Universal Declaration of Human Rights of the

United Nations, that the right to public information is a fundamental human right4.

The FOI law is of central importance as a tool for exercising democracy in Bra-

zil. It not only gives access to information previously unavailable but also requires

governments to make available a number of categories of online information and

access to open data. In fact, Law 12.527/11 is one of only a few in the world to incor-

porate several open data principles5.

1 Law 12.527 became effective only the following year in 2012. The law is known in Portuguese as “Lei de Acesso à Informação” or LAI.

2 Series C, Number 151.

3 Series C, Number 27.

4 Article 19 states: “Everyone has the right to freedom of opinion and expression; this right includes freedom to hold opinions without interference and to seek, receive, and impart information and ideas through any media and regardless of frontiers.”

5 Art. 8º, § 3º, sections II and III of Law 12.527/11 recommend that public bodies use open formats processable by computer to provide information as part of their active transparency initiatives.

introduction



Open data is an international concept developed in recent years, which re-

fers to “data that can be freely used, reused and redistributed by anyone”6. When

applied to government information the concept favors transparency by allowing

citizens to access, reuse and freely share public information7. It also maximizes po-

tential participation and public oversight of government actions.

While the law is a boon to citizens the requirements from public servants are

undeniably onerous. It establishes a new administrative paradigm, requiring public

servants to act with conscientious professionalism, competence, efficiency and pro-

bity. Compliance with the FOI law depends on an effective commitment to public

transparency. The leaders of Brazil’s public entities have a professional obligation to

comply with the FOI law, ensuring that the resultant benefits, such as more efficient

public management and greater control of the population on public spending and

policies, are spread through all spheres of government.

This study, which was presented at the seminar ‘National Evaluation on

Governmental Transparency’, represents our first effort to assess the workings

of the FOI law in practice8. The study is part of the institutional mission of the

Transparency Audit Network, an international initiative still in its developmen-

tal phase, coordinated by the Public Transparency Program at EBAPE and CTS/

FGV DIREITO RIO. The goal of the Transparency Audit Network is to develop and

apply transparency assessments in order to help civil society, governments, and

academics, to better understand and compare compliance with transparency laws

across countries and jurisdictions.

It is important to note that Brazil is recognized as a world leader in the area

of transparency, and is a founding member of the Open Government Partnership

(OGP). In this context, the country has made a commitment to promote open data,

creating a National Open Data Infrastructure (Infraestrutura Nacional de Dados

6 Open Knowledge Foundation, “Manual dos dados abertos: governo”. Laboratório Brasileiro de Cultura Di-gital & NICBr, 2011. Available at http://www.w3c.br/pub/Materiais/ PublicacoesW3C/Manual_Dados_Abertos_WEB.pdf [Accessed on: 02.11.2014].

7 For further details regarding the principles of open data see BEGHIN, N. & ZIGONI, C. (orgs.), “Avaliando os websites de transparência orçamentária nacionais e sub-nacionais e medindo impactos de dados abertos sobre direitos humanos no Brasil”. Brasília: Instituto de Estudos Socioeconômicos, 2014. Available at: http://www.inesc.org.br/biblioteca/publicacoes/textos/pesquisa-transparencia-orcamentaria-nos-websites-nacionais-e-sub-nacionais [Accessed on: 29.10.2014]

8 In Portuguese: “Avaliação Nacional de Transparência Governamental”. November 10-11, 2014, FGV, Rio de Janeiro.

21INTRODUCTION

Abertos or INDA) under the Ministry of Planning which is responsible for developing

mechanisms and standards in order to provide public information in open formats9 10.

What is the degree of compliance with the FOI law by the Brazilian Government?

In de jure terms, Brazil has one of the strongest FOI laws according to the Ri-

ght-to-Information (RTI) Rating, an international measure of the legal strength of FOI

laws11. Law 12.527/11 establishes fundamental rules to ensure citizens find information

on government websites (active transparency), and can request information not pre-

viously made available by the government (passive transparency). The Brazilian FOI

law applies to all public entities that are members of the direct administration of the

executive, legislative, judiciary and autonomous bodies12, as well as public companies,

joint stock companies and other entities of indirect public administration.

This study presents the findings of an assessment of the level of compliance

with the FOI law by public entities. The assessment is comprised of two related au-

dits: the General Audit (Part I) and the Judicial Branch Audit (Part 2). The General

Audit consisted of 453 FOI requests based on 55 individual questions. The requests

were submitted to 133 public entities across all branches of governments in the states

of São Paulo, Minas Gerais, and Rio de Janeiro, as well as their respective capitals, in

addition to the Federal District and other federal branches of government.

For the Judicial Branch Audit, 264 FOI requests were issued based on seven

specific requests. These were submitted to 40 courts: the 26 State Courts (Tribu-

nais de Justiça dos Estados), the Court of Justice of the Federal District and Territo-

ries (Tribunal de Justiça do Distrito Federal e Territórios), the five Federal Regional

Courts (Tribunais Regionais Federais), five Regional Labor Courts (Tribunais Regio-

nais do Trabalho), the Federal Supreme Court (Supremo Tribunal Federal), the Supe-

rior Court of Justice (Superior Tribunal de Justiça), and the Superior Court of Labor

(Tribunal Superior do Trabalho).

9 http://governoaberto.cgu.gov.br/index.asp

10 http://www.governoeletronico.gov.br/acoes-e-projetos/Dados-Abertos/inda-infraestrutura-nacional-de-dados-abertos

11 For more information visit: http://www.rti-rating.org/ [Accessed on: 05.11.2014]

12 For purposes of this study, Autonomous Bodies include Public Prosecutors (ministérios públicos), Public Defenders (defensorias públicas) and State Audit Institutions (tribunais de contas).



About the results

The results of the two audits demonstrate substantial differences in complian-

ce with FOI law 12.527/11. Some public entities, such as the federal government, the

executive government of the Federal District, and the executive power of the State

and Municipality of São Paulo, have adopted guidelines aimed at codifying the rules

established by the FOI law. Others however, such as the executive branches of the

State of Minas Gerais and the Municipality of Belo Horizonte still have a long way to

go towards fully implementing and complying with the requirements of the LAI, whi-

le the State and Municipality of Rio de Janeiro are significantly behind national and

international standards.

The findings suggest important progress has been made by specific federal

bodies, government branches and entities. However, other entities have introduced

barriers to transparency. We observed, for example, instances where some jurisdic-

tions issued decrees making the process of obtaining information more complex than

necessary; contrary to both the spirit and objectives of the FOI law.

Methodology

As previously discussed, this study consists of two audits: the General Audit

and the Judicial Branch Audit. Both are based on a simulated user methodology

which has been widely used in other transparency audits and is considered one of

the most effective methods of assessing a public entity’s compliance with FOI laws.

This methodology allows for the replication of a citizen’s experience, from the bure-

aucratic process of requesting public information to obtaining a response and even

appealing. The methodology was developed to be easily understood and replicable

by other institutions or persons.

The method of analysis involves the submission of information requests and the

assessment of responses, based on use three metrics:

1) response rates (number of responses received/number of requests submit-

ted); 2) accuracy rates (number of requests to receive directly relevant responses/

number of requests submitted); and 3) average response times (number of days

between the date the request was submitted and the date of response). A full des-

cription of the methodology is presented in Chapter 1 of the report.

23INTRODUCTION

Organization of the Report

This report is presented in two parts; Part I is the General Audit and Part 2 is the

Judicial Branch Audit.

Part I is set out as follows: Chapter 1 provides details about the methodology

used in both audits; Chapter 2 presents the findings of the General Audit, Chapter 3

provides examples of some of the best answers received and presents a comparati-

ve analysis of various public entities, as well as discussing barriers to transparency;

and Chapter 4 presents the findings of the assessment of digital platforms used to

manage FOI requests.

In Part 2: Chapter 5 discusses the differences in methodology between the Ge-

neral Audit and the Judicial Branch Audit, and presents the findings of the Judicial

Branch Audit; Chapter 6 discusses the barriers to transparency identified in the Bra-

zilian courts; and Chapter 7 is an assessment of the digital platforms used to manage

FOI requests. Finally, we present the overarching conclusions and recommendations

of the analysis.



1. General Audit

As described in the introduction, the Public Transparency Program (PTP), co-

ordinated by FGV, EBAPE and CTS FGV DIREITO RIO, conducted a General Audit to

assess compliance with Brazil’s new Freedom of Information Law 12.527/11. The audit

consisted of 453 FOI requests, submitted in the states of São Paulo, Minas Gerais,

and Rio de Janeiro, as well as their respective capitals, in addition to the Federal

District and the federal branches of government1.

The four states and three capitals included in this study account for nearly half

of the Brazilian population and more than half of the Gross Domestic Product (GDP).

We selected these jurisdictions due to their size and bureaucratic capacity to comply

with FOI requests. The sample is also representative of the three levels of the Brazi-

lian federal system – federal, state and municipal – as well as a variety of institutional

legacies. Finally, the sample included the federal branches of central government,

wielding the country’s largest budget, and whose policies affect all Brazilian citizens.

The selection of jurisdictions and government bodies was therefore based on

relevance within the national context. Our hypothesis was that if Law 12.527/11 is not

being properly enforced by the selected jurisdictions, its enforcement is likely to be

even more precarious in those parts of the country with more limited bureaucratic

capacity, public resources, and institutional legacies. Moreover, it is essential the law

be properly enforced within these critical jurisdictions in order to promote emula-

tion and diffuse best practice to other states and municipalities2.

The FOI requests submitted as part of the General Audit were based on three

principal themes: (i) governance and watchdogs; (ii) public participation and human

rights; and (iii) socioeconomic rights. These themes were selected for their impor-

1 For a discussion on the diffusion and adoption of public policies see: WEYLAND, K. G. Learning from Foreign Models of Policy Reform. Baltimore: John Hopkins University.

2 Ibid.

chApter 1 | methodology



tance in the current public debate and represent some of the specific interests of

professors and researchers at EBAPE FGV and CTS FGV DIREITO RIO. These sub-

jects also respond to the requests of various NGOs contacted by the Public Trans-

parency Program (PTP).

The table below shows the subject of each of the 55 FOI requests, as well as

the total number of requests submitted for each.

Table 1 – Subjects of requests and number of submissions for each request

27METHODOLOGY

It should be noted that the General Audit included six requests that had to be

submitted more than 37 times each. The other 51 requests were submitted between

one and 16 times, to different entities from the three branches of government and

autonomous institutions3.

The General Audit was developed with the objective of being comprehensive

in terms of entities and subjects analyzed, rather than an in-depth analysis of a more

limited group.

1.1. Setting up the field experiment; user identities to test for bias

The provision of a public service and the fulfillment of rights such as access

to public information are incompatible with any kind of discrimination; the provi-

sion of information should be impartial and never discriminate among applicants.

3 The complete results are available at http://transparencyaudit.net



We sought to test this principle and identify whether there is a bias in providing

access to information in Brazil. We used institutional affiliation as a proxy indicator

for class, socioeconomic status, and educational level, among other variables. Four

volunteers, who granted access and use of their personal data and were aware of the

nature and content of the requests, submitted a total of 322 requests. Two of these

identities (a man and a woman) were researchers from FGV EBAPE and CTS FGV DI-

REITO RIO. The two other identities (also one male, one female) had no institutional

affiliation or any other identification based on Internet searches.

The configuration of two pairs of requesters, male and female, institutional and

non-institutional, makes it possible to observe whether government representatives

respond differently to requests from different types of citizens.

The practice of searching for a FOI applicant’s identity on the Internet creates

opportunities for violations of the principle of equality in the provision of a public servi-

ce. As this report will show, there was evidence to suggest that the identity of applicants

was being screened4. The findings of the General Audit are presented in Chapter 2.

2. The Judicial Branch Audit

In the General Audit, the judiciary was the government branch to receive the

least freedom of information (FOI) requests. In compensation, and to provide an in-

depth analysis of one branch of government, the Judicial Branch Audit included 264

requests to courts across the country. While this audit focused on a smaller set of

topics, it covered a wider range of judicial entities.

A total of 40 Brazilian courts received between four and seven FOI requests.

The difference in the number of requests submitted to each court was due to websi-

te crashes during the submission process or because one of the question topics was

incompatible with the superior courts included in the analysis5. The majority of the

40 courts each received an average of seven FOI requests. Only one court (TJ-AC)

received four requests; the minimum number submitted.

4 In the case of profiles with institutional affiliations, it was possible to ascertain that Internet searches of their names were conducted after making the request. This investigation was conducted through tools available in institutional social networks, which notify users of this type of search. In one case (LinkedIn), one is able to see the profile of the person who made the search.

5 Federal Supreme Court (Supremo Tribunal Federal), Superior Court of Justice (Superior Tribunal de Justi-ça), and the Superior Court of Labor (Tribunal Superior do Trabalho).

29METHODOLOGY

The purpose of submitting the requests was to provide an overview of judicial

compliance with Law 12.527/11 and to obtain material information about key issues

concerning the internal management of the judicial branch.

As previously stated, both audits sought to replicate the experience of regular

citizens. Responses received in the Judicial Branch Audit were analyzed based on

the same three metrics as the General Audit: response rate, accuracy rate, and ave-

rage response time.

The topics selected for requests to the Judicial Branch included:

Compensation and pay of judges

Criteria for the promotion of judges

Mechanisms to prevent nepotism

Court budgets

Details regarding new lawsuits filed

3. Preparations; procedural rules to ensure realistic responses

All members of the research team received training on rules and procedures

regarding the submission of requests and the recording of responses. To test the

preparation of the research team, a pilot evaluation, not included in this report, was

conducted with 37 requests.

For each information request, a special form was filled out containing informa-

tion about the process, such as date of submission, site used for filing the request,

protocol number, difficulties encountered, etc6. The form also included a field to

record general observations to be used as part of a targeted qualitative analysis. All

information contained in these forms was synchronized in real-time with a database

and verified by the study coordinators7.

The submission of requests was carried out without informing the government

agency that a transparency audit was underway. All requests were submitted with

the same wording, with any differences minimal, such as presentation and signature.

6 The form can be accessed at https://docs.google.com/forms/ d/1W3vUlB9hAivtEUzCQp80-xzE8VNU7cM-NpNJaA0--pLg/viewform?embedded=true

7 A database in Google Spreadsheet format was used for these purposes.



The study adopted the following procedural rules:

In order not to alert the entity that a transparency audit was under way,

requests were submitted during different time-periods over multiple weeks.

General Audit requests were submitted during March and April 2014, while

requests regarding the Judicial Branch Audit were submitted between August

and October 2014.

We instructed each team member responsible for submitting requests, to look

for the virtual means to do so. In some cases, although the organ had a specific

digital platform for receiving FOI requests, the team could not find it because

of navigation difficulties. In such cases we used alternative mechanisms (e.g.,

ombudsman’s sites, ‘Contact Us’ pages, or general entity email addresses).

Personalized email accounts were created for each user identity and moni-

tored to track responses to information requests. Moreover, keeping in mind that

some entities did not submit responses by email, it was necessary to monitor the

passive transparency platforms used by the entities to handle requests.

Finally, two researchers evaluated and analyzed the responses. In the Gene-

ral Audit, 85 appeals were filed following the receipt of unsatisfactory responses.

Some 42 appeals were filed in the Judicial Branch Audit for the same reason. The

email accounts of the simulated users and the passive transparency platforms were

no longer monitored after July 1, 2014. Therefore, responses received after this date,

were not included in the study. A similar approach was followed in the Judicial

Branch Audit, where monitoring of email accounts and passive transparency pla-

tforms ceased November 1, 2014.

4. Metrics

This section details the three principal metrics used in both evaluations: respon-

se rates, accuracy rates and average response times. These three metrics allowed us

to analyze primary aspects of compliance with the FOI law by public bodies.

4.1. Defining what constitutes a response

The response rate is a basic indicator of the commitment of a public body to

31METHODOLOGY

its obligations under FOI law. After all, if the government does not even respond to

a request, any further transparency analysis is unfeasible.

The study’s definition of what constituted a response was broad and com-

prehensive; any kind of communication sent via email or accessible through a pla-

tform that was not an automatically generated response, was not a response about

a deadline extension, nor was related to the internal processing of the request8.

Under this definition, rejections or responses about the unavailability of information

are considered valid responses.

4.2. Evaluating response accuracy rates

In order to develop a definition of what constitutes an accurate response,

the Public Transparency Program (PTP) developed a specific methodology in three

successive phases:

1. Defining accuracy evaluation criteria.

2. Coding of all the responses received based on these criteria.

3. Calculating the results and inter-coder reliability statistics.

4.2.1. Accuracy evaluation criteria

Although our original goal was to calculate the completeness, clarity and ac-

curacy of each response, the first two proved too difficult to assess in practice, gi-

ven the wide diversity of 55 different requests and the incongruent inter-subjective

perceptions of coders. Coding decisions applied to the metrics of completeness

and clarity, simply proved statistically unreliable. This was not the case for accuracy,

however, which was the metric included in our final results.

The assessment of accuracy identified whether the response answered the

question or could be minimally correlated with the object of the FOI request. This

definition can be illustrated with the following example. If we requested a list of

all compensation paid monthly to employees, and we received the internal rules

governing the remuneration of employees, this response would not be considered

8 For example, a confirmation email to acknowledge receipt.



accurate. After all, in this hypothetical case, the agency’s response appears to eva-

de the question, instead addressing a related theme.

Additionally, the response was not considered accurate when it: (i) transferred

the request to another body; (ii) rejected the request, or (iii) communicated that

information did not exist.

4.2.2. Double coding of responses

To evaluate accuracy, two researchers coded each of the responses indepen-

dently and assigned them scores i.e., double coding procedure. The process occur-

red in two stages.

First, points were awarded to each response to determine whether it was co-

ded as fully accurate, moderately accurate or inaccurate. This calculation used a

simple ordinal scale described below:

A three-point ordinal scale was selected because it discriminates between dif-

ferent degrees of accuracy without having to make fine-grained distinctions. A scale

higher than three would be excessively detailed for a coder who would then have to

Accurate response (two points)The response is directly related to the FOI request

Minimally accurate response (one point)The response is not directly related to the request; however, thecontent is sufficiently relevant to be evaluated

Inaccurate response (zero points)The response is not related to the request, which precludes its assess-ment based on the established criteria

classify a wide variety of questions. The three-point scale also avoids the pitfalls of

a dichotomous scale, which cannot differentiate between degrees of accuracy. The

second stage of evaluation was to calculate the arithmetic mean of accuracy scores

for each response, and the average score of different entities, powers and jurisdic-

33METHODOLOGY

tions. It is important to note that in order to be considered minimally accurate, an

answer had to receive a minimum score of one from both coders.

4.2.3. Testing the reliability of double coding

The objective of using double coding was to increase the reliability of the re-

sults, since it reduces individual subjectivity. To test the reliability of the double

coding procedure, we used the kappa coefficient; a statistical measure of inter-rater

or inter-coder agreement. The kappa coefficient measures the difference in the ob-

served rate of agreement between two coders and the expected rate of agreement

if coding had occurred randomly9.

The kappa coefficient ranges between zero and one. A coefficient of one me-

ans complete agreement: a coefficient of zero means complete disagreement. The

following scale by Landis & Koch (1977) is used to assess the significance of coeffi-

cients between zero and one10.

Table 2 – Kappa scores scale

Result Agreement

0,00 Poor

0,00 a 0,20 Small

0,21 a 0,40 Moderately Small

0,41 a 0,60 Moderate

0,61 a 0,80 Substantial

0,81 a 1,00 Virtually Perfect

The results of the kappa analysis of the General Audit double coding procedure

are as follows:

9 VIEIRA, A. J., & Garrett, J. M. (2005). Understanding interobserver agreement: the kappa statistic. Family Medicine, 37(5), 360-363.

10 LANDIS, J. R., and G. G. Koch. (1977). The measurement of observer agreement for categorical data. Bio-metrics 33: 159–174.



Table 3 – Kappa score of the General Audit

Variable N Kappa Std. Dev. Agreement Expected Agreement P-value

Accuracy 315 0.6952 0.0438 89,98% 57,30% 0,000

The coding of the accuracy scores shows a high level of agreement, with a kappa coefficient of 0.6952 and a rate of agreement of 87%, well above 80%, the benchmark of excellence in the field of content analysis.

The results of the kappa analysis of the Judicial Branch Audit’s double co-ding procedure are as follows:

Table 4 – Kappa score of the Judicial Branch Audit

Variable N Kappa Std. Dev. Agreement Expected Agreement P-value

Accuracy 157 0.8537 0.0636 91,72% 43,39% 0,000

In this case, the kappa coefficient was 0.8537, an almost perfect level agreement,

confirming the study’s coding procedure as highly consistent and reliable, i.e., that is

was unlikely coding had been skewed by the subjective individual views of the coders.

4.3. Timeliness of responses

Compliance with the legal timeline, or timeliness of the response, is assessed

with a simple count of the number of days between submission of the request, and

receipt of a response. This metric reflects the commitment of the entity to comply

with the timeline established by Law 12.527/11. Paragraphs one and two of Article 11

of Law 12.527/11 state that FOI requests must be addressed by the entities within 20

days, plus a ten-day extension if necessary.

The PTP used identical methodologies in both the General and Judicial Branch

Audits to measure compliance with the FOI law. However, in addition this study pro-

vided qualitative indicators of compliance and commitments to the LAI. Qualitative

indicators include examples of positive and negative responses and an analysis of

the procedures that citizens must follow to make an information request.

1. Introduction

This chapter presents the results of the General Audit conducted by the PTP,

and coordinated by FGV EBAPE and the CTS at FGV DIREITO RIO. The results are

shown in two formats: stacked Venn diagrams and timeline graphs.

The stacked Venn diagrams summarize the following information:

Number of submitted requests;

Number and percentage of requests which received a response (response rate);

Number and percentage of requests which received accurate responses (accu-

racy rate).

The timeline graphs summarize the average number of days it took to receive a

response from a particular entity. In order to facilitate comparison, these graphs also

include information on the response rate of the respective entity, since receiving quick

responses does not mean much if the majority of requests are not answered at all.

2. Aggregate analysis

The General Audit consisted of 453 Freedom of Information (FOI) requests sub-

mitted to the legislative, executive, and judicial branches of government in the states

of São Paulo, Minas Gerais, and Rio de Janeiro, as well as their respective capitals, in

addition to the Federal District and the federal government.

As shown in Figure 1, the overall response rate of the General Audit was 69%.

This means that 31% of requests were not even acknowledged by the assessed entities.

chApter 2 | results



Figure 1 – Aggregate Results

Figure 2 below, shows the average response time for entities analyzed in the

General Audit was 21 days.

Figure 2 – Aggregate analysis of average response time

2.1. Jurisdictional level analysis

Figures three to eight present results aggregated at the jurisdictional level i.e., fede-

ral, state, and municipal, regardless of government branch or entity function. For example,

the federal government encompasses entities operating within the legislative1, executive2,

and judiciary3 powers of government, as well as autonomous entities4.

1 Federal Senate (Senado Federal) and House of Representatives (Câmara dos Deputados).

2 A diverse variety of federal government ministries, Regulating agencies, and other entities of indirect public administration (administração pública indireta).

3 Federal Supreme Court (Supremo Tribunal Federal) and Superior Court of Justice (Superior Tribunal de Justiça).

4 Federal Auditor’s Office (Tribunal de Contas da União) and Federal Public Prosecutor’s Office (Ministério Público Federal).

37

Figure 3 shows the response rate for federal entities was 83% with 76% accu-

racy. In addition to having a higher than average response rate, federal entities were

also faster in responding to the FOI requests with an average of 18 days compared

to the overall average of 21 days (Figure 4).

The results of the three states and the Federal District show significant va-

riability (Figure 5). On the one hand, the State of São Paulo responded to 80% of

requests with 73% accuracy, only the federal level has a higher accuracy rate. On the

other hand, the State of Rio de Janeiro responded to only 38% of requests submit-

ted, of which less than half (18%) were assessed as accurate.

Figure 3 – Federal level results

Figure 4 – Average response time at the federal level

RESULTS



The response rate of entities in the Federal District was the second highest in

the study after the federal level. However, the accuracy rate was much lower than

São Paulo and the federal level, and closer to the accuracy rate observed in Minas

Gerais, where more than 25% of all FOI requests were not answered, and accurate

answers represented only 61% of the total.

Figure 5 – State level results

In Figure 6, the results suggest an inverse relationship between response rates

and timeliness: the states that answered fewer requests responded more quickly.

On the other hand, the State of São Paulo and the Federal District, which had the

highest rates of response and accuracy, required a longer period than the statutory

20 days – but still responded well below the 30-day time limit.

39RESULTS

Figure 6 – Average response time at the state level

Figure 7 shows the results of the three evaluated municipalities. The City of Rio

de Janeiro, repeating the trend found in the State of Rio de Janeiro, had the lowest

response rate in the General Audit. Only 27% of all requests were answered, with just

17% answered accurately. In contrast, the City of São Paulo had the highest response

rate, with 80% of requests answered and an accuracy rate of 62%, a result slightly

higher than that obtained by the City of Belo Horizonte. There, the response rate

was 63% and the accuracy rate was 55%.



Figure 7 – Municipal level results

Regarding the timelines of responses (Figure 8), the City of Rio de Janeiro

shows a pattern similar to state level, where faster response periods correspond to

lower response rates. The City of São Paulo, however, did not follow this trend, with

the highest response rate of the three municipalities, as well as also being relatively

fast to respond (19 days on average). Belo Horizonte answered, on average, much

more slowly than the standard 21 days but within the cap stipulated by the FOI law

(30 days). This result suggests that many responses from Belo Horizonte missed the

deadlines established by Law 12.527/11.

41RESULTS

Figure 8 – Average response time at the municipal level

2.2. Analysis by branch of government

Figure 9 shows the results of the analysis by branch of government. These

results include autonomous bodies such as prosecutors, audit courts and public de-

fenders. Note the executive received more requests than any of the other powers;

almost five times that of the judiciary. For this reason we decided to conduct a spe-

cific assessment of the executive branch.

Altogether, response rates for the various branches differed by only eight per-

centage points. The judiciary, followed by the executive, presented the highest res-

ponse rates. The highest accuracy levels were achieved by the legislative and judicial

branches, at just below three-fifths of responses received; a relatively low level of

accuracy in terms of this study. In summary, all three branches and autonomous bo-

dies had very similar accuracy rates.



Figure 9 – Results by branch of government

Figure 10 presents the average response times for the different branches of go-

vernment. The judiciary responded fastest, with an average response time of 13 days,

considerably below the overall average, 21 days. The executive and legislative branches

responded in a period slightly above average, but still below the limit of 30 days.

Figure 10 – Average response time by branch of government

43RESULTS

Figure 11 provides an analysis of the performance of the executive branch di-

saggregated by jurisdictional level. The executive was the only branch to receive

a minimum number of requests (239) which allowed this type of comparison. The

results of the state and federal governments are presented below. This is followed by

the results of municipal governments compared to Federal Government.

The executive branch of the Federal District had an extremely high response

rate, but with limited accuracy. The executive branch of the federal government had

response and accuracy rates 10% higher than those of the executive branches in the

states (São Paulo, Minas Gerais and Rio de Janeiro). Again, the State of Rio de Ja-

neiro is clearly an extreme case, with a response rate of 34%, and only slightly more

than half of those, deemed accurate. Minas Gerais had 67% of requests answered of

which 57% were assessed as accurate.

Figure 11 – Analysis of executive branches by jurisdictional level: states and federal government



Figure 12 shows the average response time for the executive powers of the

three states, the Federal District and the federal government. Despite the state of

São Paulo having relatively high response and accuracy rates, it took, on average,

the time limit of 30 days. Again, this indicates that responses to some of the re-

quests did not comply with the timeline established by Law 12.527/11. The Federal

District responded, on average, in 28 days, slightly below the time limit while Minas

Gerais’ responses took an average of 25 days. It is remarkable that all of the afore-

mentioned entities took longer to respond than the aggregate average of 21 days. As

before, the results for Rio de Janeiro demonstrate an inverse relationship between

response rate and timeliness, i.e., fewer responses are answered more quickly.

Figure 12 – Average response time of executive branches by jurisdictional level: states and federal government

As shown in Figure 13, the executive branches of the municipal jurisdictions

present wide variations, with almost 40 percentage points between the worst per-

45RESULTS

forming jurisdiction (the City of Rio de Janeiro) and a jurisdiction with a moderate

performance (Belo Horizonte); and more than 20 points between a jurisdiction with

a moderate performance and the best performing jurisdiction, São Paulo. Clearly,

the Municipality of Rio de Janeiro is well behind the other municipalities regarding

compliance with the FOI law.

After the federal executive branch (included in the analysis for comparison) São

Paulo displays the best response rate, although its accuracy rate is relatively low.

Figure 13 – Analysis of executive branches by jurisdictional level: municipalities and federal government

Figure 14 illustrates the average response time for the municipal executive

branches. The executive branch of the City of São Paulo responded with remarkable

speed; well below the overall average of 21 days. The cities of Belo Horizonte and Rio

de Janeiro, on average, took longer than the 20-day limit stipulated in the law, but

remained within the 30-day limit allowed with extensions.



Figure 14 – Average response time of executive branches by jurisdictional level: municipalities and federal government

3. The field experiment; user identities and question groups

In order to test the hypothesis that institutional and non-institutional users mi-

ght be treated differently, the PTP conducted an experiment in which 322 requests

were submitted under two different user identity groups. The first consisted of one

male and one female volunteer, for whom there were no easily identifiable affilia-

tions to any institution. The second group, also a male and a female, were individuals

with easily verifiable affiliations (via Internet search), to FGV EBAPE or CTS at FGV

DIREITO RIO.

The questions were submitted under two different groups: Question Group 1

(QG1) and Question Group 2 (QG2). Table 5 shows the distribution of requests be-

tween institutional and non-institutional identities.

47RESULTS

Table 5 – Distribution of requests by user identities

QuestionGroup

No. ofrequests

Institutional identity

Non-institutional identity

Experiment with female identities QG1 154 77 77

Experiment with

male identitiesQG2 168 99 69

Total 322 176 146

Statistical tests were performed separately for each group, to test the differen-

ce in responses between institutional and non-institutional user identities.

The research design took into account the need to distribute the type of user

requests evenly across jurisdictions, so that variations in results across jurisdictions

did not influence the results. The distribution of requests by user identity and juris-

diction are presented in Table 6.

Table 6 – Distribution of requests between instititutional and non-institutional identity users

Jurisdiction Institutional identities Non-institutional identities

Federal 50% 50%

State (DF, MG, RJ, SP) 55% 45%

Municipalities (BH, SP, RJ) 56% 44%

Total 55% 45%



3.1. Results; descriptive statistics

Figure 15 presents the results by user identity, institutional vs. non-institutional,

and by gender. The overall audit response rate was 69%. The average response rate

of the institutional identity was significantly higher at 74%. On the other hand, the

aggregate non-institutional identity had a response rate of only 64.5%, almost 10%

below the institutional identity and more than 5% below the average for the entire

study. It is also important to highlight that the average accuracy rate was 51% for

non-institutional identities against 59% for institutional identities.

Figure 15 – Field experiment results

The low response rate for non-institutional, female identities is remarkable. The

non-institutional female user identity received 15% fewer answers than the institutio-

nal profile and almost 15% less than her male counterpart.

Figure 16 shows the response time for each user identity. Institutional users re-

ceived responses, on average, in 18 days. For non-institutional users that figure rose

to 26; eight days more than the institutional identities and four more days than the

aggregate study average (21 days).

49RESULTS

Figure 16 – Average response time for field experiment

3.2. Statistical tests

We used a chi-square test to determine independence between the two qua-

litative variables being analyzed. Possible differences in sample size were factored

into the test, to ensure this would not affect the result. In the case of quantitative

variables (average days for response) we used a t-test for comparison of means

between groups.

Table 7 – Chi-square test for differences in response rate

QuestionGroup N Institutional

identity (%)

Non-institutionalidentity (%)

Chi-square

Significance (p-value)

Experiment with female identities QG1 154 70,10% 57% 2,8 0,094*

Experiment with male identities QG2 168 74,70% 72,50% 0,11 0,740

(*):p<0,1



Given the results in Table 7, the experiment with QG1 supports the hypothesis

that institutional and non-institutional users are treated differently; specifically that

response rates might be better for institutional identities than for their non-insti-

tutional equivalents. The test for QG1 was significant at a confidence level of 90%,

which rejects the null hypothesis (equality between the variables). However, the ex-

periment with QG2 could not confirm independence between the two variables,

given that the result is not statistically significant.

Table 8 – Chi-square test of differences in accuracy rates

QuestionGroup N Institutional

identity (%)Non-institutionalidentity (%)

Chi-square


Accuracy score 0 1 2 0 1 2

Experiment with female identities

QG1 98 15% 17% 69% 20,4% 18,1% 61,3% 0,667 0,717

Experiment with male identities

QG2 124 20% 9% 70% 20% 14% 66% 0,625 0,731

In order to conduct the chi-square test with accuracy as the dependent variable,

it was necessary to add categories to the analysis5. Table 8 presents the test results.

The chi-square test did not show statistically significant results for either ques-

tion group. This leads to a rejection of the hypothesis that accuracy rates would be

higher for institutional than for non-institutional identities i.e., there is no evidence

that an identity with an institutional affiliation influences the accuracy of the response.

We used the average number of days for response from each question group,

to test for difference in response times between institutional and non-institutional

profiles, and these results are shown in Table 9 below.

5 Among the possible scores of this variable, we added the requests with a score of 0.5 to requests with a score of 1. We also added those requests with a score of 1.5 to the observations with a score of 2. In this way, the possible classification categories of the accuracy variable, was reduced to 3 (i.e., the accuracy variable may take the values 0, 1 or 2).

51RESULTS

Table 9 – T-test of differences in average response time

QuestionGroup N

Average response time in days (institutional)

Average response time in days (non– institutional)


Experiment with female identities QG1 98 17 24 0,065*

Experiment with male identities QG2 124 18 27 0,015**

(*):p<0,1(**):p<0,05

4. Conclusions

To summarize, the results of the field experiment suggest the current rules on

access to public information promote discrimination in the treatment of applicants.

Non-institutional users were found to receive fewer responses (almost 10% less), on

average eight days later.

Table 10 – Results of hypothesis testing

Institutional affiliation affects response rate?

Institutional affiliation affects accuracy rate?

Institutional affiliation affects timeliness?

Female Identities (QG1) Confirmed within the sample

Not confirmed withinthe sample

Confirmed within the sample

Male Identities (QG2) Not confirmed withinthe sample



The results support the general hypothesis of discrimination in terms of res-

ponse rates and timeliness of responses. Of particular interest, results show that wo-

men with an institutional identity received responses, on average, seven days before

women with no institutional identity: men with an institutional identity received res-

ponses to requests nine days ahead of those without. This infers a prioritization of

requests within the public entities based on the institutional status of the applicant.

The influence of institutional affiliation on the accuracy of responses provided

by public agencies cannot be affirmed in statistical terms. However, the lack of sta-

tistically significant results might suggest that once a public entity decides to res-



pond to an information request it does so impartially, regardless of the individual’s

institutional affiliation.

It is important to highlight that reducing opportunities for discrimination is

possible with the adoption of recognized best practice, in Mexico and Chile for

example, where systems explicitly do not require the submission of a citizen’s iden-

tification card to exercise the fundamental right of access to information: contact

details are deemed sufficient.

Obliging official identification introduces risks of discrimination and intimida-

tion, particularly at local level, where staff can easily identify citizens. Preventing dis-

crimination and intimidation is all the more important in those places most in need

of freedom of information, for example in areas where criminal organizations have

infiltrated government, and where citizens are already likely to pose fewer inquiries

about the administration of government.

5. Appeals; their place in the study

5.1. Introduction; background and procedure

We filed 85 appeals as part of the study. Appeals were filed for a random

sample of information requests in cases where it was deemed appropriate by our

analysts. It is important to note that other FOI law audits rarely include the submis-

sion of appeals because they require significant additional work.

Article 15 of Law 12.527/11 stipulates an appeal might be made if the entity

rejects or denies the request, or if the quality of response does not meet the user’s

expectations. The deadline for filing an appeal is ten days following the original

response. A citizen then has the right to a second appeal, addressed to a higher

authority. According to Article 16 of Law 12.527/11, third round appeals must be filed

with the Comptroller General of the Union. However, this option is only available in

cases where the body rejecting the second appeal belongs to the federal govern-

ment. Among the 85 appeals submitted, 78 were first round appeals and seven were

second round: no third round appeals were filed.

53RESULTS

Those cases which resulted in appeals included situations where the public entity:

claimed the request was improperly filed, generic or not appropriate; and as

such, the entity stated, it was unable to provide an answer;

demanded the request be submitted in person, as was the case with several

executive bodies in the City and state of Rio de Janeiro;

requested the response be retrieved in person at its headquarters;

complained of additional work necessary to compile the information requested;

unjustifiably claimed secrecy;

responded with a generic or incomplete response;

sent, in response, a web-link that did not contain the requested information.

5.2. Successful appeals; overturning unsatisfactory responses

In numerous cases, the appeals were an important mechanism in obtaining

a more accurate response than the initial one. This section presents a few of the

successful appeals filed, which reinforce the idea that the appeal process is a useful

mechanism to overturn an unsatisfactory initial response to a FOI request.

For example, we requested from the State of São Paulo’s Department of Edu-

cation, a spreadsheet containing a list of all contracts exempt from the bidding pro-

cess, as well as the values of each contract and the justification for the exemption.

We obtained, as a first response, only a generic text on the procurement procedures

within the entity. However, after we filed an appeal, the agency provided a detailed

table listing all contracts exempt from the bidding process in 2013, in addition to the

purpose of the contract and its value.

The PTP observed the same positive effect of an appeal in relation to a re-

quest sent to the National Transportation Agency (Agência Nacional de Transportes

Terrestres or ANTT). The request asked for information regarding all fines imposed

on concessionaries regulated by the agency, e.g., values, number of petitions and

justifications. In its first response, ANTT claimed the requested information was con-

fidential based on an internal resolution of the agency6. However, after an appeal

6 Resolution ANTT number 442 from 2004, which regulates the FOI law within the agency.



in which we explained that the said resolution did not apply to the subject of the

request, the agency sent tables containing information on the fines imposed on each

concessionary, their value, and explanations broken down by mode of transport (rail,

road and terrestrial).

Lastly, a similar case occurred with a request sent to the São Paulo State Uni-

versity (Universidade Estadual Paulista or UNESP) for information on their integra-

ted program for the development of partnerships with the private sector. We also

requested information on the measurement of the university’s impact on local and

regional development. In the first response, UNESP provided an evasive and generic

answer. However, following an appeal, the organization provided information with

precise and detailed explanations on their processes, and the measurement of im-

pact on local and regional development.

1. Introduction

The findings presented in the previous chapter provide an estimate of the insti-

tutional compliance with the FOI law. However, the content of the answers received

can provide additional insights about successful compliance and the existence of bar-

riers in implementing the law. For this reason, we present an analysis of (i) the recei-

ved responses and best practices, and (ii) observed trends in barriers to transparency.

2. Best practices

The cases presented below illustrate how the FOI law can be a powerful tool

in helping civil society find relevant public information. Some of the best practices

encountered by this study are as follows:

The Program for Public Transparency at the FGV (PTP/FGV) submitted a

request to the Auditor’s Office of the State of São Paulo (Tribunal de Contas

do Estado de São Paulo or TCE-SP) for lists of all contracts exempted from the

bidding process in 2013. The TCE-SP submitted a spreadsheet with reasons

to support each waiver, e.g., well recognized legal experience, price, etc., as

defined by Law 8.666/93.

The PTP/FGV submitted a request to the Court of the State of Rio de Janeiro

(TJ-RJ) for information the court already had in its active transparency webpage:

consolidated and organized information about contracts exempted from the bidding

process. By bringing transparency to these contracts, the TJ-RJ has adopted a

model useful to other organs. In addition to providing the value, the reason for the

contract, and the justification for waiving the bidding process, it allowed access to

the entire content of the contract. Photocopies of all executed contracts exempt

from bidding are made available for consultation on its active transparency website.

chApter 3Best prActices And BArriers to trAnspArency



The PTP/FGV asked the Ministry of Education for the total amount it had

disbursed and what it had been charged, for work undertaken on the construction

and contracting of day-care facilities by different municipalities. The ministry

submitted a spreadsheet containing information on works of approximately 9,300

day-care facilities in different municipalities. The spreadsheet included details

relating to (i) the total amount committed to each project, (ii) the amount that

had been actually paid, and (iii) the percentage of work already completed. The

sum of the values committed to daycare facilities throughout Brazil totaled R$

9,604,666,930: the magnitude of this value shows the relevance of the topic and

hence the need for public administrators to have information properly organized.

The PTP/FGV issued a request to the House of Representatives for the 2013

parliamentary plenary sessions’ attendance roll. The response included specific

links to the web page of the active transparency portal of the organ, containing

a spreadsheet with the information requested in a format able to be processed

by computer.

3. Barriers to Public Transparency

In addition to best practices, this study also identified the opposite, i.e., those

practices creating serious obstacles to government transparency and to the effecti-

ve functioning of Law 12.527/11. Such cases included:

Requiring that the request be submitted in person or that the answer be sought

in person, contradicting Article 10, paragraph 2, of the FOI law;

Rejecting the request based on broad interpretation of the ‘additional work’ clause;

Transmitting information in formats not able to be processed by computers;

Aggregation of information at a different level to that requested;

Implementing potentially intimidating procedures, such as necessitating the

signing of liability forms, for the use of information.

The study found the following examples of the above listed practices, which

might undermine or limit freedom of information.

57BEST PRACTICES AND BARRIERS TO TRANSPARENCY

3.1. Requirements that the request be submitted or retrieved in person

Some public agencies rejected several requests for information submitted by

digital means. Many time times this justification was based on a specific regulation.

For example, the State of Rio de Janeiro issued Decree 43.597/12, which covers

entities within the executive branch, and which stipulates that information requests

must be submitted in person.

However, the FOI law 12.527/11 does not require that information requests be

submitted in person. In fact, the law stipulates that FOI requests to federal, state or

local government agencies may be submitted by “any legitimate means”. Moreover,

the FOI law actively encourages the use of digital means for the submission of re-

quests. Article 3, III, states that procedures for handling information requests must

be compatible with the use of; “communication means made available by technolo-

gical developments.”

Therefore, the requirement by the State of Rio de Janeiro is illegal and goes

beyond the regulatory powers of the executive branch, making the procedure of

access to information more restrictive than stipulated by federal law.

For these reasons, the State of Rio de Janeiro’s decree represents a real obs-

tacle to the freedom of information in the state1. The requirements stipulated in the

decree can completely preclude the exercise of this right, contrary to the objectives

of the FOI law and the provisions of Article 5, XXXIII, of the federal Constitution.

Unfortunately, we encountered barriers, similar to the ones found in the execu-

tive branch of the State of Rio de Janeiro, in the following entities:

Auditor’s Office (Tribunal de Contas) of the State of Rio de Janeiro2

Department of Education (Secretaria de Educação) of the State of Rio de Janeiro3 4

1 Article 8 of Decree 43.597/12 reads as follows: “The interested party must submit a request, in the standard form (Appendix I) along with the statement of responsibility (Appendix II), to the agency or entity that has the required documents.”

2 Request confirmation number (número de protocolo): 149.040.051.441 and 146.010.352.485.

3 The refusal by this entity is not supported by Municipal Decree 35.606 of 2012, which in art. 8, discusses the possibility of submitting a request via the Internet. We note, nevertheless, refusal by the Department of Educa-tion of Rio de Janeiro on the grounds of requiring submission of the request in person.

4 Request confirmation number (número de protocolo): RIO-5902239-4.



In other cases, the entity assessed accepted the request sent digitally, but in-

dicated that the answer should be sought in person at its headquarters. This requi-

rement also represents an obstacle to the law, which does not require citizens to be

present for a response. It is clear that such a requirement could dramatically reduce

the effectiveness of the law, especially (but not only) in cases where the applicants

reside in other cities or states. The identification of barriers imposed by the govern-

ments of the State and City of Rio de Janeiro leads to the conclusion that both have

adopted measures contradictory to the spirit of Law 12.527/11.

The requirement that responses to FOI requests be retrieved in person, was

found in the following entities:

Legislative Assembly (Assembleia Legislativa) of the State of São Paulo5

Court of Justice (Tribunal de Justiça) of the Federal District6

Legislative Assembly (Câmara Legislativa) of the Federal District7

The Federal Public Prosecutor’s Office (Ministério Público Federal)8

Department of Transportation (Secretaria de Transporte) of Belo Horizonte9

3.2. Rejection of the request based on a broad interpretation of the ‘undue additional work’ clause

Several FOI requests were rejected in the first instance, based on the claim that

producing the answer would require additional work in the analysis, interpretation,

or gathering of data and information.

There is a normative basis for public bodies to use the clause of ‘additional

work’ to reject an access to information request: for example Article 13, III, of federal

Decree 7.724 of 2012 which regulates Law 12.527/11 at the federal level, in addition to

5 Request confirmation number (número de protocolo): 645

6 Request confirmation number (número de protocolo): 2014-068123

7 Request confirmation number (número de protocolo): 62/2014-OUV


9 A request confirmation number was not provided by the entity.


numerous state and municipal decrees10. However, several of the entities analyzed

used these provisions too broadly, denying legitimate claims, in violation of the FOI

law. The same problem was identified by the NGO, Article 19, in its Annual Report to

the Regional Alliance for Freedom and Expression and Information (Alianza Regio-

nal por la Libre Expresión and Información)(2014)11.

It should be noted that it is intrinsic to the FOI requesting process that some

degree of additional work will be required of public agencies, especially in cases

where the entity has not conducted prior work organizing and consolidating re-

quested information. After all, it is common that the information requested, despite

addressing basic issues from the point of view of efficient management of a public

body, has not been previously produced.

The additional work claim is particularly unjustified in instances when the FOI

request concerns items in Section § 1, Article 8, of Law 12.527/11, which stipulates an

obligation to actively disseminate and disclose information, regardless of citizens’

requests: it might be considered a blatant abuse of the additional work clause given

that the information requested should already exist independent of the access to

information request12.

We noted this additional work claim was used to deny access to documents of

significant public interest, for example, by the Health Department of the Municipality

of São Paulo13. We submitted a request to this agency to provide a spreadsheet con-

taining all contracts entered by the agency with a bidding-requirement exemption

in 2013. It was expected that this information had already been compiled, as it is

required by section VI of Art. 7, § 3, of Law 12.527/11.

10 Art. 13 – “The following FOI requests will not be accepted: III – requests which require additional work for the analysis, interpretation, or gathering of data and information, or production services or formatting of data which is outside of the competencies of the entity.”

11 ALIANZA REGIONAL POR LA LIBRE EXPRESIÓN E INFORMACIÓN. Saber más VI: informe regional sobre acceso a información pública y apelaciones (2014).

12 “Art.8 It is the duty of public bodies and entities to promote, independently of requirements, the dissemi-nation of information of collective interest that is produced by or under the oversight of the public body. Infor-mation must be disseminated through an easy to access medium, and within the entity’s area of competency. § 1st Information to be disseminated as part of compliance with Art. 8, must include, at minimum: I – record of competence and organizational structure, addresses and telephone numbers of their units and opening hours to the public; II – records of any transfers of financial resources; III – records of expenditures; IV – information concerning bidding procedures, including the respective notices and results, as well as all contracts; V – general data for the monitoring of programs, activities, projects and works of agencies and entities; and VI – answers to frequently asked questions.”

13 Request confirmation number (número de protocolo): 8476.



A similar case occurred with the Department of Education in the City of São

Paulo in response to a request for a list of those individuals who had failed the proba-

tion period as municipal teachers14. Initially the department denied the request clai-

ming confidentiality. Following an appeal, the department denied the request again,

on the grounds that to do so would generate additional work in gathering informa-

tion. Based on this response, it can be inferred that neither the Department of Educa-

tion nor the Health Department of São Paulo had the requested data readily availab-

le; there would be no reason to decline the requests otherwise. A negative response

can only suggest deficient management, if not poor data and archival practices.

These experiences raise an important point: it is essential that public agencies

adopt information management mechanisms to ensure the required information is

easily accessible. An appropriate archival practice becomes particularly important

when related to key information such as public expenditure, procurement, transfers

of financial resources, and employee figures, for example.

In addition to the Departments of Health and Education of the State of São Pau-

lo, the following entities also declined requests based on the ‘additional work’ clause:

The Lower House of the National Legislature (Câmara dos Deputados)15

Legislative Assembly (Assembleia Legislativa) of the State of Minas Gerais16 17

Federal Communications Commission (Ministério das Comunicações)18

Department of Education (Secretaria de Educação) of the State of Minas Gerais19

3.3. Transmitting information in formats not processable by computers

Some of the FOI requests asked for the information to be formatted in a spre-

adsheet. This was the case in requests submitted to the entity for all consultancy

service contracts and spending on advertising in the previous three years.

14 Request confirmation number (número de protocolo): 8539.

15 Art. 7, VI of the Ato da Mesa nº 45/2012 which regulates Law 12.527/11 in House of Representatives includes the the “additional work” clause.

16 The “additional work” clause is based on Decision (Deliberação) 2.555/2013, which regulates the implemen-tation of FOI law within the Legislative Assembly of the State of Minas Gerais.


18 Request confirmation number (número de protocolo): 53850.001261/2014-14

19 Request confirmation number (número de protocolo): El9Vm7kZ


As stated earlier, it is crucial that government data is made available to citizens

in computable format; that it be “structured to enable its automated processing”

thereby allowing applicants, should they so wish, to use and analyze the data in a

different way20.

In fact, the Brazilian legislature, when drafting Law 12.527/11, was concerned

with addressing this specific issue, as stated in Article 8, § 3, sections II and III. There

is a clear intention in the law, to promote formats processable by computer, as well

as open formats and non-proprietary formats.

For the purposes of the General Audit, we considered the following formats

as computer processable: csv, xls, ods, doc, docx and odt2122. The PDF (Portable

Document Format) was considered processable when it was possible to copy the

contents, and unprocessable when it was not.

We received various responses in non-processable form, as shown in the list below:

Federal Auditor’s Office (Tribunal de Contas da União)23

Auditor’s Office (Tribunal de Contas) of the State of São Paulo24

Lower House of the National Legislature (Câmara dos Deputados)25

City Council (Câmara de Vereadores) of Belo Horizonte26

20 BEGHIN, N. & ZIGONI, C. (orgs), “Avaliando os websites de transparência orçamentária nacionais e sub-na-cionais e medindo impactos de dados abertos sobre direitos humanos no Brasil”. Brasília: Instituto de Estudos Socioeconômicos, 2014. Available at: http://www.inesc.org.br/biblioteca/publicacoes/textos/pesquisa-transpa-renciaorcamentaria-nos-websites-nacionais-e-sub-nacionais [Accessed on: 29.10.2014]

21 This is not an exhaustive list. There are many other formats that can be considered processable or readable by computers. It should be noted, however, that only files in proprietary formats were received during this research.

22 As discussed in the introduction, in order to be considered open format, the data must meet multiple re-quirements. Automated processing is only one of such requirements. Also, the use of non-proprietary formats is essential for the dissemination of data. For that reason, proprietary formats are not considered in this study. For a more in-depth analysis about compliance with open data principles see the already mentioned study by BEGHIN & ZIGONI (2014).



25 Request confirmation number (número de protocolo): F186103136367




3.4. Aggregation of information in a pattern different from the original request

Some information is only relevant and useful when aggregated or disaggrega-

ted at the level requested. For example, we submitted requests to various legislative

assemblies and city councils for individualized attendance lists pertaining to mem-

bers of the assembly and council in 2013 plenary meetings. The request, however,

expressly requested that information be aggregated annually, so that it could be

possible to rank representatives based on their attendance for each year.

However, instead of submitting the information annually as requested, some

entities submitted it disaggregated monthly. The answer, therefore, had no real value

because the objective comparison of the presence of different elected representa-

tives was not feasible27. Consequently, the answer did not allow us to infer which

parliamentarians had higher and lower rates of attendance at plenary sessions; in-

formation of clear public interest.

It was interesting to note that by sending the response with a different level of

aggregation than requested, the agency complied with Law 12.527/11 in appearance,

but in reality imposed an obstacle to transparency.

This problem was encountered with the following entities:

City Council (Câmara de Vereadores) of Rio de Janeiro28

Legislative Assembly (Assembleia Legislativa) of the State of São Paulo29

Legislative Assembly (Assembleia Legislativa) of the State of Minas Gerais30

3.5. Implementation of potentially intimidating procedures

In addition to requiring the filing of information requests in person, the exe-

cutive branch of the State of Rio de Janeiro requires that petitioners complete a

‘Statement of Responsibility for Use and Disclosure of Information’31. This statement

27 A herculean effort would be need to aggregate the data manually

28 A request confirmation was not provided by the entity.



31 “Termo de Responsabilidade pelo Uso e Divulgação da Informação” http://www.rj.gov. br/web/informa-caopublica/exibeconteudo?article-id=1038866


stipulates the person making the request accepts liability for damages arising out of

possible ‘misuse’ of the information received, citing the possibility of criminal libel,

slander and defamation charges.

Such requirements might generate practices of self-censorship, especially for cit-

izens without legal knowledge, i.e., the vast majority of the population in Brazil. We, the

authors, believe such practices are severely detrimental to an individual’s willingness to

use the freedom of information law, and on the freedom of expression itself. In addition

to the executive branch of the State of Rio de Janeiro, this practice was also identified

in the Auditor’s Office (Tribunal de Contas) of the State of Rio de Janeiro32.

A similar practice, although less severe, was observed in the passive transpar-

ency platform of the Public Ministry of the State of Minas Gerais (Ministério Público

de Minas Gerais), which required citizens, before sending a request for information,

to select the following option:

“I declare awareness of, and acceptance that, in accordance with fed-

eral Law 12.527 of November 18, 2011, and Resolution 89/2012 of the National

Public Prosecutor’s Council [Conselho Nacional do Ministério Público], in-

appropriate use of the information obtained from the Public Ministry of the

State of Minas Gerais, will be subject to or responsible for legal penalties.”33

In this specific case, it is emphasized that the contents of Res. 89/2012 of the

National Public Prosecution Council are not made available, which makes it impos-

sible for citizens to become aware as to what they are actually agreeing. As in the

case of Rio de Janeiro, the above content might potentially discourage some citi-

zens from seeking or disseminating public information. Also, this type of warning is

unnecessary, since illegal practices can clearly result in legal penalties, regardless of

the use of information obtained.

32 The requirement of signing the “Statement of Responsibility for Use and Disclosure of Information” is esta-blished in Resolution Number 275 of 2013 of the Auditor’s Office of the State of Rio de Janeiro.

33 http://www.mpmg.mp.br/acesso-a-informacao/requerimento-de-informacao/



1. Introduction

Researchers from the FGV’s Program for Public Transparency (PTP/ FGV)

found great disparity between public entities in terms of the quality and existence

of digital platforms for making and receiving FOI requests. This chapter presents the

analysis results of 33 digital platforms used for this purpose.

For a FOI law to work, it is important the process for sending requests and

receiving responses be as simple as possible. Therefore it is essential that govern-

ments adopt digital platforms, which are both efficient and uncomplicated. Ideally

a platform should also provides a search tool for citizens to identify responses

to similar requests, avoiding costly duplications: Mexico, for example, has such

a system, and many other countries are in process of adopting or implementing

such frameworks.

The existence of a digital platform dedicated to FOI requests simplifies a pro-

cess that some citizens regard as complicated, scaffolding their inquiry, and since

requests are recorded the possibility that a request can be ignored is reduced.

In Brazil, while the access to information system, e-SIC, used by the federal

government has a number of attributes that facilitate the process, other entities

have adopted platforms or systems which impose barriers.

2. Information access barriers: a lack of compliant digital platforms

Some Brazilian public entities do not even have a platform specifically desig-

ned for receiving FOI requests. Such entities often transfer this task to their om-

budsman’s site, or ‘Contact Us’ webpage. This option causes some problems, which

we describe in the following sections.

chApter 4Audit of the plAtforms used to mAnAge foi requets



2.1. Impossibility of filing an appeal

Law 12.527/11 provides specific guidelines relating to the process of accessing

information, for which an ombudsman’s website might prove inadequate. An exam-

ple is the right to appeal a decision when a FOI request is rejected. Since the om-

budsman’s website has no specifically designed tool for filing appeals, the citizen is

obliged to file through the same channel that the original, rejected FOI law request,

was submitted. This can confuse the user, who may then fail to file an appeal at all,

unable to find a channel they deem appropriate.

The filing of appeals can play an important role, not only for the user who

wants to access specific information, but also for the government. The existence of

appeals may indicate, for example, that the department responsible for complying

with FOI requests is not operating properly. Thus, tools to facilitate the submission

of appeals might help public bodies improve management of resources, for the ef-

fective enforcement of freedom of information.

2.2. Character limits

Another example of problems arising from the use of the ombudsman’s we-

bsite is the character limit. The existence of a limit in the field where users request

information sets a clear obstacle to transparency and is a violation of Law 12.527/11,

which does not establish any rules as to the length of requests.

3. Evaluation of digital platforms

The FOI law stipulates that public entities should provide an “option on their

official Internet websites through which to submit requests.”1 It does not define,

however, the specific features such digital platforms must have, leaving this to the

entity’s discretion.

With this in mind, the PTP conducted a specific assessment of 33 government

entities’ websites through which information requests were submitted. The goal was

to identify the characteristics of the best platforms, as well as obstacles presented

by those platforms, or ombudsmen’s websites, and ‘Contact Us’ webpages, which

proved unsatisfactory.

1 Article 10, paragraph 2 of Law 12.527/11

67AUDIT OF THE PLATFORMS USED TO MANAGE FOI REQUETS

3.1. Methodology of evaluation; what constitutes a good FOI digital platform?

Through the analysis of several different platforms around the world it is possible

to identify the main features a good FOI platform must possess. Some of these are pre-

sent on the e-SIC platform created by Brazil’s federal government. Despite not having

the functionality to allow searches of all requests sent by any and all applicants, and

their respective answers, e-SIC is considered a good example of a passive transparency

platform, and should be seen as a model for other governmental entities in the country.

Positive attributes of e-SIC include the following characteristics:

The user registers in the system which grants access to a platform on which

all requests sent and responses received are recorded. When an information

request is submitted, the system emails an automatic confirmation of receipt;

The system sends an automatic email notification when the request is answered;

The platform has a specific section for the filing of appeals;

The use of a log-in procedure means that access to the answer is possible

regardless of the availability of the confirmation or protocol number.

The following sections discuss in detail the importance of these attributes.

3.2. Submission of a confirmation email

Following submission of an access to information request, an email confirming

receipt is of great importance; it reassures the user that the body has received their

request, and that the FOI process is actively underway.

3.3. Submission of a notification of response

Sending an email stating that the access to information request has been answe-

red allows citizens to become more quickly aware of the information’s availability.

3.4. Availability of a specific section for filing appeals

As previously discussed, the ability to file appeals in cases where the answer

is unsatisfactory is an important right of citizens, granted by law. Various platforms,



however, do not have a specific section for this purpose. Consequently, the citizen is

obliged to appeal through the same channel used to request access to information:

potentially confusing and discouraging, since it is unclear in the agency’s platform

whether the right to an appeal exists.

3.5. Use of log-in procedure instead of confirmation/protocol numbers

Several of the platforms audited only allowed citizens to access responses to

the requests after inserting the confirmation or protocol number generated when

the original request was submitted. These platforms did not offer citizens an alterna-

tive mechanism to enable access to the contents of the response in case of the loss

of the confirmation/protocol number.

The lack of a log-in feature does not violate the parameters of the FOI law. Howe-

ver, it does make the platform less user-friendly and can pose a problem if the appli-

cant, for any reason, does not properly record the number of confirmation/protocol.

3.6. Summary

Given these barriers to access, the PTP/FGV advocates a digital-platform whi-

ch (i) unifies into a single log-in procedure all information requests and subsequent

responses for a user; and which (ii) prevents instances where the loss of a protocol

number means the impossibility of accessing the response.

4. Comparative analysis of digital platforms

The data contained in the table below refers to the digital platforms of all 33

entities analyzed.

69

Table 11 – Platforms evaluation

AUDIT OF THE PLATFORMS USED TO MANAGE FOI REQUETS



It should be noted that a single platform is available to submit FOI requests to

the executive branches at the federal, state and municipal levels. Therefore the indi-

vidual platforms of each entity, i.e., ministries and departments, were not analyzed.

The reader should also take into account that the assessment of digital platforms

ended on June 30, 2014.

The first column of Table 11 identifies whether or not the entity analyzed has a

specific digital platform for the submissions of FOI requests. The next four columns

refer to desirable attributes of a good passive transparency digital platform.

5. Conclusions

Taking into account that freedom of information is a constitutionally protected

right, it is important that requests based on Law 12.527/11 be handled through dedi-

cated platforms, rather than lumped together with a variety of requests unrelated to

FOI, in a platform used for other purposes.

Furthermore, it is recommended that agencies conform to the technological

realities and adopt digital platforms to manage FOI requests. Mexico, Chile, parts of

India, and Canada, all provide examples of international best practice in this regard.

As previously noted, it is also ideal than any digital platform used to handle

FOI requests features a global search engine to permit the search of previous FOI

requests. This would help agencies avoid needless and costly replication of labor at

the taxpayer’s expense.

In contrast to the executive and legislative branches, members of the judiciary

are not elected and cannot be removed through a vote. The tenure of judges serves

as an important mechanism in promoting their independence but also introduces

risks. At one extreme, the judiciary may be captured by special interests while at the

other, the judiciary may become so independent, that it is scarcely accountable to

the polity. For these reasons, democracy experts caution that the development of

accountability mechanisms beginning with greater transparency is critical to ensur-

ing the integrity and legitimacy of the judicial branch of government.

The Program for Public Transparency (PTP/FGV), coordinated by FGV EBAPE

and the CTS at FGV DIREITO RIO, conducted an audit of the Brazilian judicial branch

with the goal of assessing its level of compliance with the FOI law1. As mentioned

in the introduction, the Judicial Branch Audit used the same methodology as the

General Audit, thus there is a high level of comparability between the general and

specific findings of the judicial branch.

This audit aims to assess the active and passive transparency of the judiciary,

through the gathering and analysis of data. Several recent studies have focused

on analyzing how the Brazilian judiciary has dealt with the right of freedom of in-

formation. These include: “Year 1 Audit of the Freedom of Information law and the

Brazilian Judiciary – 2012-2013” by the NGO Article 19, as well as “Transparency

Challenges in the Brazilian Justice System” by the Secretariat of Judicial Reform

(Secretaria de Reforma do Judiciário) in collaboration with Article 19, FGV CEAPG

1 Two other recent projects by the FGV DIREITO RIO used data analysis to understand and illuminate from different angles the problems of the Brazilian justice system. In the analysis Supremo in Numbers a large volume of data on nearly 1.5 million cases of the Supreme Court (Supremo Tribunal Federal) is analyzed from different angles, such as the variety of cases, the geographical distribution, the duration of procedural steps, and so on. Project results can be found in http://www.fgv.br/supremoemnumeros/ [Accessed on: 30.10.2014]

pArt ii | JudiciAl BrAnch Audit

By Luiz Fernando Moncau, raFaeL VeLasco, GreGory Michener, Marina Barros, and JaMiLa Venturini



and Gpopai/USP2 3 4 5. The current study seeks to add new dimensions to this research.

2 “Balanço de 1 Ano da Lei de Acesso à Informação Pública: acesso à informação e os órgãos de justiça bra-sileiros – 2012-2013”

3 “Desafios da Transparência no Sistema de Justiça Brasileiro”

4 Center for the Study of Public Administration and Government (Centro de Estudos em Administração Pú-blica e Governo)

5 Public Policy Research Group/University of São Paulo (Grupo de Pesquisa em Políticas Públicas/University of São Paulo)

1. Methodology

The methodology used in the Judicial Branch Audit replicated the methodolo-

gy used in the General Audit, explained in detail previously in Chapter 1.

We adjusted the methodology in superficial ways to the peculiarities of the ju-

diciary, as well as to meet the scope and objectives of the investigation. The details

of these adjustments are described below.

1.1. Audit coverage

This audit assessed 40 Brazilian courts, including Superior Courts, Regional La-

bor Courts, Federal Courts, and Courts of Justice. Table 12 below presents distribution.

Table 12 – Distribution of audited courts

Courts Quantity

Superior CourtsSupreme Court

(STF), STJ and TST3

Regional Labor Courts TRTs 5*

Regional Federal Courts TRFs 5**

State Courts (Tribunais de Justiça) TJs 27***

Total 40

(*)TRT of the 1st Region, TRT of the 2nd Region, TRT of the 3rd Region, TRT of the 10th Region and the TRT of the 15th Region.(**)TRFs of the 1st and 5th Regions.(***)TJs of all the Brazilian states and the Federal District

chApter 5 | results



Seven FOI requests were submitted to each court, making a total of 280.

However, one of the requests did not apply to the superior courts, reducing the total

number of requests by three. Additionally, 13 requests were canceled for technical

reasons. Ultimately, 264 FOI requests were included in the audit.

The following is a list of the 16 requests not included in the audit:

Three requests about the promotion of judges (Request Category 4) not

were submitted because they did not apply to the Superior Courts (Tribunais

Superiores Supreme Court (STF), STJ, and TST);

Two requests with incorrect questions were excluded from the analysis;

Eleven requests were not submitted due to problems with the websites for the

following tribunals: TJ-BA (two requests), TJ-AM (one request), TJ-AC (three

requests), TJ-RS (two requests), STJ (one request), TJ-RO (one request), and

Supreme Court (STF) (one request). The exclusion of these requests did not

impact the assessment of these courts.

1.2. Development and submission of requests

Seven request types were developed and submitted to the courts included in

the audit. The topics addressed by the requests are shown in Table 13.

1.3. Topics of requests

We attempted to frame each request in a clear and concise manner in order to

leave little room for interpretation by the entity, thereby reducing the possibilities of

a rejection. Thus, legal terminology was used in the development of the questions

whenever necessary.

The objective of requests 1, 2, and 7, was to obtain information about the com-

pensation of judges. The so-called National Administrative Reform (Reforma Admi-

nistrativa do Estado), implemented by Constitutional Amendment Number 19/98,

established a ceiling for the compensation of public servants. This constitutional

amendment is enforced through a complex web of laws and regulations, which in

addition to the constitutional norms, includes the Internal Laws of the National Ju-

75RESULTS

diciary (Lei Orgânica da Magistratura Nacional or LOMAN), numerous laws, and Re-

solutions 13 and 14, 2006, of the National Judicial Council (Conselho Nacional de

Justiça – CNJ).1 2 3 4

Table 13 – Request Topics of the Judicial Branch Audit

Requests Topic

Request 1 Remuneration for the years 2012 and 2013

Request 2Remuneration for the years 2012 and 2013, with

additional details

Request 3 Nepotism

Request 4 Career Plans and promotions

Request 5 Budget for the years 2012 and 2013

Request 6Number of cases judged in the first instance,

for the years 2012 and 2013

Request 7* Remuneration, elaborating on Request #1

(*) Request 7 was used very specifically, as detailed below.

The legal complexities tackled in requests 1, 2, and 7, become even more bewil-

dering given the intricacies of compensation schemes devised for judicial civil ser-

vants and judges.

It is accepted as essential to promote greater transparency in the remuneration

structure of the judiciary, a task originally attempted through the adoption of the

National Judicial Council (CNJ) Resolution No. 102, of December 15, 2009, almost

1 See for example Article 37, section X to XII, Article 39, § 4, and Article 93, section V.

2 See for example Articles 61 to 77, which address salaries, benefits and other advantages of the profession, as well as Article 32, which addresses the irreducibility of judges’ salaries. The complete law is available at: http://www.planalto.gov.br/ccivil_03/ leis/lcp/lcp35.htm.

3 See for example Law No. 8.350 of 1991, which sets the value of bonuses for presence in election of justices (Justiça Eleitoral), and Law No. 12.771/2012, which effectively fixes the salaries of the Supreme Court (STF) judges.

4 Resolutions 13 and 14 address, respectively, “the application of a constitutionally mandated compensation ceiling for members of the bench” and “the application of a constitutionally mandated compensation ceiling for public servants of the judicial branch and of the state courts that do not receive a subsidy”.



two years before the adoption of the FOI law.

Resolution 102, amended by Resolution No. 151, 2012, requires the publication

of not only the compensation structure of the courts and councils, but also other pa-

rameters regarding budgeting and staffing structures5. The resolution requires the

publication of a table containing information on the base salary of judges, as well

as personal benefits, allowances and other discretionary benefits judges may earn.

However, these categories are not sufficient to clarify under what rubrics judges are

receiving different types of payments. For this reason, FOI requests were formulated

to allow an ordinary citizen to gather more detailed information on the remuneration

structure of the judiciary in all of the evaluated courts.

As discussed below, we formulated Request Categories 1 and 2 considering

this legal and regulatory framework, using the technical terms of the National Judi-

cial Council (CNJ)’s own tables.

Request Category 1 requested the following information:

Consistent with the provisions of the annex of Resolution No. 151 of the Natio-

nal Judicial Council (Conselho Nacional de Justiça – CNJ), I request the amounts

paid to each judge in the form of base salary, personal benefits, subsidies, allowa-

nces and other discretionary benefits. Please provide this information in a compu-

ter-processable format, following the guidelines established by the National Judicial

Council (CNJ), summarized on an annual basis for the years 2012 and 2013. The table

should provide the total annual amount of all compensation categories paid to each

judge of this Court during the years 2012 and 2013.

With this request, we sought to obtain data already disclosed by the courts in

their electronic sites pursuant to Resolution No. 102/2006, as amended by Resolu-

tion No. 151/2012 of the National Judicial Council (CNJ). However, the information on

the website is often not aggregated by year, and, as such, does not permit citizens

to understand the total annual compensation for each judge. Also, data is often not

in a computer-processable format, i.e., it is not possible to perform an automated

analysis, without for example, a document containing a spreadsheet. Moreover, the

5 National Judicial Council (CNJ)’s Resolution No. 151 amends Article 3, section VI of Resolution No 102. This amendment was essential since it establishes a new model for the dissemination of information regarding the “salary, per diem, allowances, and any other monies paid to members of the bench and other public servants under any title…” But more than that, the amendment also forced the judiciary entities to disseminate, via their websites, not only pay, but also the name of the judges. This led to a completely new level of transparency on salaries within the judiciary.

77RESULTS

data often cannot be obtained in a single query; multiple queries frequently have to

be done on the judge’s name, for example.

Request Category 2 asked for the following information from the courts:

(a.1) What are all the categories of allowances, as defined in Article 3, section VI

and Annex VIII of Resolution 102, as amended by Resolution 151 of the National Judicial

Council (CNJ), that public servants and judges of this court can receive? Please sum-

marize, by occupation title and by compensation category, the criteria for receiving

such allowances, as well as the maximum monthly amount that judges can receive.

(a.2) For each type of allowance (e.g., meals, transportation, preschool, health

care, child-birth assistance, housekeeping, relocation, and other allowances of this

nature, as answered in section a.1), we request information on amounts that were paid

to each judge (identified by name) in each of the months of 2012 and 2013.

(b.1) What are all the categories of personal benefits, as defined by Resolution

151 of the National Judicial Council (CNJ), that public servants and judges of this

court can receive? Please summarize, by occupation title and benefit category, the

criteria for receiving such personal benefits, as well as the maximum amount that

can be paid monthly.

(b.2) For each type of personal benefit, e.g., V.P.N.I.[6], extra payment for length

of service [7], and retention bonuses[8], in addition to any other personal benefits

consistent with the response to section (b.1), I request information regarding the

amount paid to each judge (identified by name) in each of the months during the

years 2012 and 2013.

(c.1) What are all the categories of discretionary benefits, as defined in Resolu-

tion 151 of the National Judicial Council (CNJ), which public servants and judges of

this court can receive? Please summarize, by occupation title and benefit category,

the criteria for receiving such personal benefits, as well as the maximum monthly

amount that can be paid.

(c.2) For each discretionary benefit, e.g., constitutional bonus of 1/3 of vaca-

tion, paid vacation, advance of vacation payment [9], Christmas bonuses, advance of

6 Personal Benefit Identifiable by Name or Vantagem Pessoal Nominalmente Identificável

7 “Adicional por tempo de serviço” in Portuguese.

8 “Abono de permanência” in Portuguese.

9 “Antecipação de férias” in Portuguese.



Christmas bonuses, overtime, substitution, retroactive payments, and other discre-

tionary benefits consistent with the response to section (c.1), we request information

on the amounts paid to each judge (identified by name) in each of the months during

the years 2012 and 2013.

(d) In addition to the categories mentioned above and set forth in Resolution 151

of the National Judicial Council (CNJ), are there any other sources of income or funds

that judges of this Court can receive?

These requests asked that the courts disaggregate and detail information on

the compensation of judges, which is already publicly available via their online por-

tals. Through these requests we sought to gain a more in-depth understanding of

the compensation structure of Brazilian judges.

Request Category 3 addressed another point of great importance for public

management as well as for the Brazilian judiciary – nepotism. The basis for this re-

quest was the National Judicial Council (CNJ) Binding Precedent (Súmula Vinculan-

te) No. 1310 of the Supreme Court (STF), as well as Resolution No. 7 from October 18,

2005. The terminology of this resolution was adopted to inquire which procedures

have been adopted to prevent nepotism by the courts evaluated. The full text of

Request Category 3 is presented here:

Based on Law 12.527, I would like to request the following information:

Does this Court adopt, in addition to the self declaration form, any other proce-

dure(s) at the time of hiring an employee for an atwill appointment [11], or at the time

of the appointment of a civil servant to a position of trust [12], to ensure the absence

of family ties with other employees of the Court, which may constitute nepotism as

defined by Resolution No. 7 of 2005 of the National Judicial Council (CNJ)?

If this Court adopts the aforementioned procedures, please submit all docu-

ments related to these procedures.

10 The Binding Precedent generally prohibits any official, or body responsible, for making a hiring/appoint-ment decision, from appointing, promoting, or recommending for appointment or promotion any relative, friend, or spouse to any agency or department over which the official or body exercises authority or control. This applies to all the branches of the federal government, the states, the federal district and municipalities. See http://www.stf.jus.br/portal/jurisprudencia/menusumario.asp?sumula=1227 [Accessed on: 27.10.11]

11 “Cargo em comissão” in Portuguese.

12 “Função de confiança” in Portuguese.

79RESULTS

Request Category 4 addressed the criteria for promotion of judges in the Bra-

zilian courts. These requests asked for:

All internal norms (e.g., ordinances, resolutions, etc.) of this Court relating to

the promotion of judges. I kindly request that the full text of these norms be sent.

(a) All meeting minutes of the relevant authorities (e.g., special committees,

Presidency, etc.) for the decision of the promotion of judges in which this matter has

been discussed.

Request Category 5 dealt with the budget, while Category 6 asked for informa-

tion on the number of new cases in the courts. Information for both is already public

and can be found in the report Justice in Numbers (2014) which is released annually

by the National Judicial Council (CNJ).13

Request Category 5 read as follows:

I would like to request the following information based on Law 12.527:

What was the budget of this court during the years 2012 and 2013? I would like

to request a detailed breakdown of monies used for the purchase of goods, contract-

ing of services, as well as the amounts used for payroll.

Request Category 6 read as follows:

Based on Law 12.527 I would like to request the following:

What was the total number of new cases filed in courts by this body during the

years 2012 and 2013?

The goal of requesting this preexisting information was to identify variations in

response rates, as well as calculating the accuracy and average response times (see

definition of these metrics in Chapter 1). It also helped us gauge commitments to

timely disclosure, as we were able to compare these responses to responses which

required greater effort to collect and systematize information.

A court’s inability to answer any of the questions in Requests 5 and 6 could in-

dicate serious problems in the implementation of the FOI law. On the other hand, if a

13 Justiça em Números. Ver: http://www.National Judicial Council (CNJ).jus.br/ programas-de-a-a-z/eficien-cia-modernizacao-e-transparencia/pj-justica-em-numeros [Accessed on : 03.11.2014]



court responded appropriately to requests 5 and 6, ignoring or responding inappro-

priately to all or some of the other requests, it might tell us something about the

court’s commitment to FOI.

As previously discussed, Request Category 1 specified that answers should be

made available in a format to allow the processing of data, as well as its possible

combination with other data.

This provision of computer-processable formats is one of the principles of

open data and is encouraged by the FOI law14. According to the report “Transparen-

cy Challenges in the Brazilian Justice System”15;

“The most effective form of creating transparency is to publish infor-

mation in open formats readable by computers, which allows for the reuse

and more efficient analysis of information.”16

A recent analysis conducted by the Socioeconomic Studies Institute (Instituto

de Estudos Socioeconômicos) and the Public Policy Research Group/University of

São Paulo (Grupo de Pesquisa em Políticas Públicas/University of São Paulo or Gpo-

pai/USP) analyzed transparency initiatives by different public entities in Brazil; with

an emphasis on open data17. Their assessment is that there is “an enormous distance

between reality and the legal requirements which stipulate the publication of bud-

geting information in an open format.”

Although our specific purpose was not to analyze compliance with the princi-

ples of open data, the results obtained in the Judicial Branch Audit in response to

14 According to the Open Knowledge Foundation, open format data is “data that can be freely used, reused and redistributed by anyone”. When applied to government information, the concept of open data encourages transparency by allowing citizens to access, reuse and freely share public information. It also creates opportu-nities for society to participate in and oversee government actions. Making government data available in open format means, among other things, the ability for automatized processing, the use of non-proprietary formats, and the absence of restrictions based on authoring, patents, branding, or trade secrets. For more information see: Open Knowledge Foundation, “Manual dos dados abertos: governo”. Laboratório Brasileiro de Cultura Digital & NICBr, 2011. Available at: http://www.w3c.br/pub/Materiais/PublicacoesW3C/Manual_Dados_ Aber-tos_WEB.pdf [Accessed on: 02.11.2014].

15 “Desafios da Transparência no Sistema de Justiça Brasileiro”

16 SAMPAIO, ALEXANDRE, et.al. “Os desafios da transparência no Judiciário brasileiro.”

17 BEGHIN, N. & ZIGONI, C. (orgs), “Avaliando os websites de transparência orçamentária nacionais e sub-na-cionais e medindo impactos de dados abertos sobre direitos humanos no Brasil”. Brasília: Instituto de Estudos Socioeconômicos, 2014. Available at: http://www.inesc.org.br/biblioteca/publicacoes/textos/pesquisa-transpa-rencia-orcamentaria-nos-websites-nacionais-e-sub-nacionais [Accessed on 29.10.2014]

81RESULTS

Request Category 1 showed the challenges are numerous and that courts have not

yet duly followed through on their obligations.

As discussed below, the prevailing interpretation is that the availability of data

in PDF (Portable Document Format) or HTML (Hypertext Markup Language) is su-

fficient to meet the obligations of active transparency in relation to the FOI law.

However, it is important to note that these formats limit the extraction of data and

are not the most suitable for the processing of data in spreadsheets. Thus, although

the information is made available, it is not in a format processable by computers.

Responses that did not comply with the specified format were therefore con-

sidered inaccurate. The reason given in many cases was that the data was already

available to the public pursuant to Resolution No. 102 of the National Judicial Cou-

ncil (CNJ). Therefore, a new wave of requests was submitted under Category 7, the

text of which read:

“I am aware that the details of the payroll of judges by month are available in

the Transparency Portal of this Court in the format stipulated by Resolution 102 of the

National Judicial Council.

In view of this fact, and based on law 12.527, I would like to request the detailed

spreadsheets of the payroll of judges for each month of the years 2012 and 2013, in a

format processable by computer, which can be .xls or .xlsx (excel), or some other for-

mat processable by computer. After all, the format used by the Transparency Portal of

this court, that is, .pdf, is not computer-processable, preventing proper data analysis.

It should be highlighted that the preparation of detailed spreadsheets of the

judges’ payroll certainly required the use of excel or similar free software, as it is

impossible to organize this data directly in a PDF document. Therefore it can be

assumed that this body possesses the data in question in the requested format and

there is no legitimate argument for refusing to supply this information.

Furthermore, sections II and III of paragraph 3 of Art. 8 of Law 12.527 reinforce

the importance that the information contained in the Transparency Portals subject to

the law are made available in open and machine-processable formats.”



1.4. User identities: simulating the request process experienced by a regular citizen

One of the goals of this study was to simulate the process experienced by ci-

tizens interested in making use of the FOI law. To accomplish this, seven volunteers

made their identities available for the submission of 264 FOI requests. These indivi-

duals were employees of FGV EBAPE and CTS at FGV DIREITO RIO. It is important

to note that there was no mention of the requesters’ affiliation to these institutions.

Only the required information, such as name, identity documents, and email address

were provided. The objective of using multiple identities was to minimize the possi-

bility that entities would realize an audit was underway, possibly skewing responses.

1.5. Submission of requests

All requests were submitted via Internet through the most relevant platform

for each entity: dedicated platforms for handling FOI requests, ombudman’s web-

sites, email or any other platform available. The requests were submitted over a

three-week period, between August and October 2014. The only rule was that a

single user could not submit two requests on the same day to the same entity. The

following table provides details of FOI requests submission.

Table 14 – Requests by identity

Timeframes

of requestsIdentities Requests

Total requests

issued

Total requests

not issued

Total requests

planned

Week 1 A/B 1 and 2 77 3 80

Week 2 C/D/G 3,4 and 7 113 7 120

Week 3 E/F 5 and 6 74 6 80

Total 264 16 280

1.6. Appeals

As mentioned in the General Audit, Article 15 of Law 12.527/11 establishes the

possibility of appealing against decisions to reject a FOI request.

While the General Audit submitted appeals for a random sample of requests

83RESULTS

in order to identify differences in the behavior of entities, the Judiciary Audit filed

42 specific appeals for requests 1, 2, 3 and 4, in order to obtain the requested data.

(Requests 5 and 6 were not subject to appeals) The appeals followed the same legal

rigor of the original requests, incorporating the appropriate legal terminology.

In summary, appeals were filed because the entity either rejected the FOI re-

quest, or the response did not provide the level of detail requested, or was not in

the format requested.

Those entities providing the best appeal responses were: the TST, TRF Region 2,

TRF Region 1, TJ-DFT and TJ-RR, the latter responding appropriately to two appeals.

1.7 Field experiment; testing responses to justified vs. non-justified requests

The use of more than one identity served a second purpose: to test whether

the evaluated organs would respond differently to justified requests versus non-jus-

tified requests.

Forcing citizens to justify their FOI requests goes against international stan-

dards of freedom of information, and Brazilian law follows best practice in this re-

gard. Article 10, § 3 of Law 12.527/11, prohibits any requirement that the user justify

their request. It is possible perhaps that justifying a request might increase the li-

kelihood of receiving a satisfactory response; the audit therefore included a field

experiment to explore the paradoxes associated with justifying the practice of a

fundamental right.

To conduct this field experiment the identities of users were divided into two

groups; one to justify its requests, i.e., clarify the purpose for which information

was sought (e.g., research); whilst the other group provided no explanations. The

purpose was to assess whether justified and non-justified requests received equal

treatment in terms of timeliness of response, and accuracy. To conduct this experi-

ment without arousing suspicion on the part of the audited agencies, we chose not

to send the same requests twice to a particular court. Instead, requests were divided

as shown in Table 15, with each entity receiving three justified, and three non-justi-

fied. The study aimed at minimizing evidence an audit was underway since it might

influence a court’s behavior and so compromise the experiment results.

The courts in Group X received requests 1, 3, and 5 with a justification, and re-

quests 2, 4, and 6 without justification. The courts in Group Y received requests 2, 4

and 6 with a justification and requests 1, 3, and 5 without.



2. Results

Table 15 – Courts by group

Group X Group Y

Supreme Court

(STF)STJ

TST TRT SP / SP Inter

TRT DF TRT MG

TRT RJ TRF 3

TRT 1 TRF 4

TRT 2 TRF 5

(+) 14 TJs (+) 13 TJs

20 20

This section presents the aggregate results of the Judicial Branch Audit. The

audit used the same metrics presented in Chapter 1: response rate, accuracy rate,

and timeliness of response.

Seven requests were submitted to each of the 40 courts, totaling 264 valid

requests. Of these, the courts responded to 160 requests, which corresponds to a

response rate of 61%. Some 69 responses (26%) were assessed as accurate. In other

words, only 26% of the courts provided directly relevant responses. It is important

to note that the FOI law establishes a 20-day period to respond to requests, plus a

10-day extension if properly justified (art. 11, § 1 and § 2).

Regarding the timeliness of the response, the judiciary responded, on average,

within 12 days of request submission.

2.1. Results by request category

The results of the audit varied according to request category: this section pre-

sents those results.

85RESULTS

Figure 17 – Aggregate analysis results of Judicial Branch Audit

Request Category 1

As previously explained, the first category of requests sought information on

the compensation of judges during the years 2012 and 2013 (based on the require-

ments of the annex of Resolution No. 151 of the National Judicial Council or CNJ).

This request asked for the compensation data, by judge and by year, in a computer-

-processable format. The substance of the request falls in line with the National Ju-

dicial Council (CNJ) rule, which stipulates that information must be made available

on court Internet sites18. However, the majority of courts’ websites do not allow for

the importation of spreadsheets with aggregate data, nor for the filtering of data by

attributes, such as month or name of judge. Given these limitations, the objective of

Request Category 1 was to obtain data, which could be aggregated and analyzed.

Thirty-nine such requests were submitted. The courts responded to 25 (64%) of the-

se; however, only five (13%) were considered accurate responses.

Those courts whose responses were assessed as accurate, submitted sprea-

dsheets, i.e., provided a file containing the compensation data which could be ag-

gregated by year and by judge. The Supreme Court (STF) and the Court of Justice,

18 The National Judicial Council’s (CNJ) Resolution No. 102, and its amending Resolution No. 151, stipulate such information be made available on the websites of the judiciary entities in HTML format (Article 4, paragraph 3), which is not a computer-processable format. Moreover, since no other guidelines are provided regarding the personnel payroll, there are disparities between the different entities’ ability to provide the values aggregated on an annual basis.



Roraima (TJ-RR), provided the data already summarized on an annual basis. The

Supreme Labor Court (TST), Federal Regional Court (TRF) Region 2, and STJ sub-

mitted spreadsheets with monthly data, which could be aggregated with computer

software to obtain the annual amounts.

Figure 18 – Results of Request Category 1

Those responses to merely direct requests to the active transparency links of

the relevant data were considered inaccurate, since the format of the data availab-

le on those platforms does not allow it to be aggregated on an annual basis. Even

within this type of response, there were differences in the quality of service given.

For example, some courts provided the specific links to the available information,

thereby demonstrating interest in providing a better service to the citizen. Other

entities, such as the Court of Justice, Paraíba (TJ-PB), only mentioned the agency’s

website without providing a direct link. It is important here to remember Article 11, §

6, of the FOI law which states that information already available to the public ‘must’

be provided by the entity, as well as stipulating how the entity might do so19.

19 Article 11 states that the public body or entity must authorize or provide immediate access to information already available. § 6 states that if the information requested is already available to the public in printed, elec-tronic or any other media of universal access, the entity must inform the person making the request, in writing, of the mode and location in which the data can be consulted, obtained or copied. This would release the public body or entity of the obligation to provide the data directly, unless the person making the requests does not have the means to complete the process by his/herself.

87RESULTS

It should be noted that some courts are still not up to date regarding the disse-

mination of data concerning compensation of judges, as stipulated by the National

Judicial Council’s (CNJ) Resolution No. 102 of 2009 and amending Resolution No.

151 of 2012. For example, according to TRT Region 10:

“[…] the Department of Information Technology and Communication

and the Office of Personnel Coordination and Functional Information [20]

are making the necessary adjustments to the personnel system with the

objective of meeting the requirements of the National Judicial Council’s

(CNJ) Resolutions No. 102 and No. 151.”

Regarding timeliness, the average response time for Request Category 1 was 14

days. This is illustrated in the following figure.

Figure 19 – Average response time for Request Category 1

20 “Secretaria de Tecnologia da Informação e Comunicações” and “Coordenadoria de Pessoal e de Informa-ções Funcionais”, respectively.



Request Category 2


The second request category also sought information regarding the compen-

sation of judges during the years 2012 and 2013, but requested additional details

regarding the different categories of allowances, personal benefits, discretionary

benefits and other income or funds. Also this request inquired about the criteria

used by each entity to grant such additional income or funds to judges, as well as

the maximum monthly amounts allowed.

Thirty-eight (38) requests were submitted within this category and 26 received

responses (68%). Of the 26 responses, only six (16%) were assessed as accurate.

Regarding timeliness, the average response time for Request Category 2 was

14 days after the submission of the request, as illustrated in the figure next page.

Request Category 2 did not specify a format for data. Responses were consi-

dered accurate solely by the entity’s act of responding with the requested data, re-

gardless of whether the format facilitated any analysis of the information provided.

The responses of the Superior Labor Court (TST) and Court of Justice, Roraima (TJ-

-RR), stood out for the substantial amount of data provided in response to requests.

89RESULTS


Some 25 days after the submission of the request, the TST provided:

Two documents in PDF format with the monthly amounts received by each

judge in the categories of allowances, personal benefits, and discretionary

benefits during the years 2012 and 2013.

An official document in PDF format, with a digital signature by the Director of

the Payroll Department of the Court, which addressed in detail each of the sub-

sections of the requests. The explanations included the relevant values and norms.

By the same token, the TJ-RR submitted two files in .zip format, which contai-

ned the documents described below. It is worth highlighting that these files were

received only four days after the request was submitted. The TJ-RR provided:

The schedule of allowances paid by the court, its legal basis, the value/percentage/

estimate basis and/or criteria/conditions for granting such allowances.

The schedule of personal benefits paid by the court, its legal basis, the value/

percentage/estimate basis and/or criteria/conditions for granting such benefits.

The schedule of discretionary benefits paid by the court, its legal basis, the value/

percentage/estimate basis and/or criteria/conditions for granting such payments.



Two tables on the discretionary benefits paid to judges monthly, for 2012 and 2013,

Two tables on the allowances paid to judges, monthly for 2012 and 2013.

The law does not require a specific format for responses to the FOI requests.

However, best practice makes use of non-proprietary formats, at the very least, pro-

cessable by computer, in addition to alignment with the open data philosophy. The-

se principles increase the chances that citizens will be able to make productive use

of the data.

Request Category 3

Request Category 3 asked for information regarding the procedures used by

the entity to prevent practices which might be considered nepotism according to

Resolution No. 7 of the National Judicial Council (CNJ)21. Thirty-seven requests were

submitted under this category: the audited entities responded to 25 (68%), and

of the responses received, 23 (62%), were evaluated as accurate. As shown in the

following figure, this response rate is slightly above the response rate for the audit

overall, at 61%.


21 Available at: http://www.National Judicial Council (CNJ).jus.br/atos-administrativos/atos-da-presidencia/323-resolucoes/12121-resolu-no-7-de-18-de-outubro-de-2005-

91RESULTS

Despite the high accuracy rate, the majority of responses did not contain docu-

ments related to the procedures. Most only mentioned the related resolutions and norms

adopted by the entity. However, the TST, TRT Region 15, TJ-MT, TJ-TO, and TJ-MA inclu-

ded the Declaration of Familial Relationships (or “termo de declaração de parentesco”).

The average response rate for Request Category 3 was 13 days, as shown in

the figure below.


Request Category 4

Request Category 4 asked each court for the complete text of the relevant

internal norms, as well as the minutes from meetings regarding deliberations per-

taining to the promotion of judges. Thirty-six requests were submitted under this

category: responses were received for 25 (or 69%), with just six (or 17%), evaluated

as accurate. Despite a response rate above the overall average of the Judicial Branch

Audit, the accuracy rate was significantly lower.




Regarding timeliness, the average response time was 17 days after the sub-

mission of the request, as illustrated in the following figure. Among those entities

providing responses evaluated as accurate were TRF Region 1, TJ-DFT, and TJ-RR22.

The TRF Region 1 first responded five days after the submission of the requests, di-

recting the user to its active transparency portal. An appeal was filed and a response

was received 19 days later, some 24 days after the initial submission of the original

request. This final response contained the web-links to all of the documents listed in

the original FOI law request.

22 For purposes of the analysis by request category, these three responses were coded as accurate since the final answer was taken into account. On the other hand, for consistency purposes, the aggregated audit results (Table 16) only take into account initial responses. Therefore, since these answers were obtained only after an appeal, they were coded as inaccurate in the aggregated audit results.

93RESULTS


The TJ-DFT provided its first response only four days after a request was sub-

mitted. However, an appeal was filed since the response was incomplete, only poin-

ting perfunctorily to the website of the court. Following the appeal, the court sub-

mitted five scanned documents containing the norms and meeting minutes related

to the promotion of judges.

The TJ-RR submitted its first response three days after the request was made.

Its final response, however, was received 38 days after the initial request, also after

the submission of an appeal. The final response by the TJ-RR included two scanned

documents in PDF format with the requested norms and meeting minutes.

Request Category 5

Request Category 5 addressed the budget of the entities for the years 2012

and 2013. We requested a detailed breakdown of monies used for the purchase of

goods, the contracting of services, and for payroll costs. Thirty-six requests were

submitted, of which 22 received responses (61%). Twelve of those responses (or

33%) were coded as accurate.

The responses received for Request Category 5 highlight the variability in res-

ponses to requests for data already published. The responses coded as accurate

were those in which: a link was provided to the documents requested; the court

submitted a spreadsheet with the requested expenditures (as in the case of TRF



Region 2); or the court detailed the expenditures in the body of the response email

as in the case of TRFs Region 4 and 15.



Regarding timeliness, the courts responded to the requests, on average within

nine days. This is well below the average response times for the Judicial Branch Au-

dit and the General Audit.

95RESULTS

Request Category 6

Request Category 6 sought information firstly regarding the quantity of new

cases filed during the years 2012 and 2013. Thirty-eight requests were submitted, of

which 23 (61%) received responses, 15 of them accurate; an accuracy rate of 39%.


Similarly to Request Category 5, the responses to Request Category 6 were

received in a short period, the average response time being six days, as shown in the

following figure.

When requests did not specify a format or a level of detail, the accuracy rate

was higher than those which did (Request Categories 1, 2, 3, 4 and 7).




We also observed that the average response time for those responses asses-

sed as accurate was well below the 20-day deadline stipulated in Article 11 of the

FOI law. It is worth highlighting the cases of TJ-DFT and TJ-MG which submitted

satisfactory answers in a single day. The TJ-DFT submitted the complete annual re-

ports in PDF format, while TJ-MG provided instructions on how to find the requested

information on the court’s website.

Request Category 7

This category also asked for information about the compensation of judges

during the years 2012 and 2013. These requests, however, asked for the information

in a computer-processable format23.

A total of 40 requests were submitted of which 14 received responses (35%),

with two responses assessed as accurate (5%).

23 These requests are based on Article 8 of Law 12.527/11 which reinforces the importance that the informa-tion contained in the transparency portals of entities subject to the law be provided in open format and are pro-cessable by computer. According to Article 8: “It is the duty of public bodies and entities to promote, regardless of requirements, the dissemination through readily accessible means, within its ability, of information of public or general interest produced by, or under custody of the agency. […] § 3 The sites addressed by § 2 must, per regulation, among others, comply with the following requirements: […] II – enable the production of reports in a variety of electronic formats, including open and non-proprietary formats, such as spreadsheets and text, with the goal of facilitating the analysis of the information; III – enable automated processing by external systems in open formats, structured and readable by computers; […]”.

97RESULTS


The audited courts were relatively fast in responding to the requests. As shown

below, the average response rate was 11 days.


We must commend the performance of the TST, which submitted its responses

within ten days. The response included two spreadsheets in a proprietary-format;

not ideal practice, but which are processable by computer (.xls). The spreadsheets

included information regarding the monthly compensation of judges, including base



salary, personal benefits, subsidy for the position of trust, allowances, and discretio-

nary benefits during the years 2012 and 2013.

Another notable response, but in a negative sense, came from the Supreme

Court (STF). It stated that the court’s transparency portal did not have the function

of exporting the data into excel or other machine-processable formats. According

to the court’s response:

“The information is exported directly from the database into a PDF for-

mat, therefore, an intermediate phase of aggregating the data is not possible.”

This is not the case we believe. The court should have still been able to comply

with the request despite not having the capacity to aggregate the data using the

court’s software, since nearly all database programs have an option to export data

into non-proprietary formats, such as a comma-delimited or tab-delimited format.

Once in this format, data could be aggregated or analyzed, by a myriad of non-pro-

prietary computer programs. Therefore, the court should have been able to export

the data requested, directly into one of these formats instead of directly into the PDF.

As has been repeatedly stated, the FOI law requires that data be provided in an

open format and be processable by computer. Therefore, the very fact that the Supre-

me Court’s (STF) software cannot export public interest data into the stipulated format

suggests the need to update the court’s system to comply with legal requirements.

2.2. Detailed results

This section presents response and accuracy rates, as well as the average response

times of the field experiment: justified requests vs. non-justified requests.

99RESULTS

Figure 32 – Field experiment results: response and accuracy rates

User identities A, C, and E did not provide explanations for their requests, i.e.,

the requests were non-justified. Identities B, D, and F justified their requests based

on academic research or professional reasons.

A total of 113 justified requests and 111 non-justified requests were submitted24.

There was no difference in the response rate between the two groups, both with

65%. The accuracy rate was also similar: 30% (34 accurate answers) for justified re-

quests vs. 31% (34 accurate responses) for non-justified requests.

24 The difference in the number of submissions under each category is the result of deviations in the research, which are explained in the methodology section of this chapter.



Figure 33 – Field experiment results: average response time

The results of the field experiment suggest the audited courts did not differen-

tiate between justified and non-justified requests. From this perspective, the courts’

practices comply with Article 10, § 3 of the FOI law, which prohibits any rules which

might oblige requesters to justify access to public information requests.”25

Nevertheless, one court did base its rejection of a request on the grounds that

it lacked “sufficient justification”; namely in the three rejections by the Court of Jus-

tice of the State of Santa Catarina (or TJ-SC)26. In its rejections, the TJ-SC referenced

a decision by the entity from 2006, in which a request was rejected because the

information “was not made available for personal use”, and because insufficient ex-

planations were provided regarding the “reasons for making the request”.

These responses illustrate one of the potential barriers to freedom of informa-

tion; a broad interpretation of exemptions included in the FOI law (outlined further

in Chapter 6, Barriers to Transparency and Access).

25 Available at: http://www.planalto.gov.br/ccivil_03/_ato2011-2014/2011/lei/l12527.htm

26 Out of seven requests submitted to the Court, the TJ-SC rejected three, did not respond to two, and res-ponded to two others inaccurately or with minimal accuracy

101RESULTS

2.3. Analysis of the courts

The following table shows a list of the audited courts, ranked by accuracy. For

consistency purposes, this table only takes into account the first response submitted

by the courts: not responses obtained after filing appeals.

The ranking table shows inadequate performance by the majority of the courts

with respect to both response and accuracy rates. The State Courts (Tribunais Esta-

duais) had an average accuracy rate of 14%, well below the total average of 24%. Mo-

reover, the TJ-AM, TJ-AP, and the TJ-PA did not respond to any of the FOI requests.

The Superior Courts (Tribunais Superiores) and the Regional Courts (Tribunais

Regionais) had the best performance27. Among the 12 top-ranked courts are the

Superior Labor Court (TST), the Supreme Court (STF), the Regional Federal Court

(TRF) Region 4, the Regional Labor Court (TRT) Region 15, and the TRT Region 1.

The TRF Region 5, TRT Region 2, and the TRF Region 1 had very low respon-

se and accuracy rates. These courts ranked 19th (both TRT Regions 5 and 2) and

22nd, respectively.

Among the state courts, two stood out for their positive performance; the al-

ready mentioned Courts of Justice TJ-RR, and the TJ-RJ, ranking among the top 12

best performing tribunals.

Regarding timeliness of response, some of the results obtained by this study

raise serious questions concerning the courts’ abilities to respond to FOI requests,

demonstrating a potential focus on legal compliance with the timeline stipulated,

rather than with providing accurate and useful responses. The cases of the TJ-RN,

TJ-BA, TJ-RS, and TRF Region 1 illustrate this point. These courts used only half of

the response period permitted by the FOI law, but achieved accuracy rates of zero.

Such practice could represent a barrier to freedom of information, indicating an

emphasis on generating any response, as opposed to one that might actually meet

the needs of the user – true access to information. It is important to note that the

FOI law permits a ten-day extension beyond the 20-day period, through a justified

request (Article 11, paragraph 2).

Overall, the average response time for most of the courts is below the 20 days

established by the law. The only courts which did not comply with this were; the TRT

Region 1, TJ-AC, and TJ-AL.

27 Including Federal (Federais) and Labor (de Trabalho) Courts.



Table 16 – Court ranking

When the PTP/FGV analyzed the responses of the Brazilian courts to the stu-

dy’s FOI requests, it became evident that many challenges remain in guaranteeing

the right to public information. This section highlights the main obstacles to achie-

ving compliance with the FOI law. Both the General and Judiciary Branch Audits

identified some key barriers and challenges. However, certain practices in the ju-

diciary merit specific attention. Here we identified a culture of excessively legalis-

tic justifications in which unfounded rationales for the denial of FOI requests were

commonplace. In many instances we observed that responses pointed to a complex

maze of legal norms, with responses so complicated that it was impossible to un-

derstand the basis for the court’s decision.

Therefore the challenge for Brazilian courts is to find a balance between res-

ponding with clarity and simplicity, while continuing to properly explain the basis for

the denial of FOI requests.

In an attempt to categorize the challenges encountered, the barriers to FOI are

divided as follows:

Bureaucratic barriers: regarding the method of submitting the request (e.g.,

specific platform, email, in-person, etc.), requirements for the request (e.g.,

required information, etc.), or selection of the appropriate entity/department

responsible for handling requests;

Legalistic, or broad interpretation of the law barriers: regarding the legal basis

for the rejection of requests, or cases in which broad interpretations of the

exemptions included in the FOI law were used;

Technical barriers: regarding the format or the level of aggregation of the

requested information, which could preclude the visualization or analysis of the

information requested.

chApter 6BArriers to trAnspArency And Access



The following sections provide a more detailed analysis of these three types

of barriers.

1. Bureaucratic barriers

Among the bureaucratic barriers identified by the evaluators, two stood out:

The response that the body or entity to which the request had been submitted was

incorrect, or that the body or entity did not possess the information requested.

The response that the request was not submitted in an appropriate format or

that the request did not comply with a formal requirement.

An example of the first bureaucratic barrier identified is the response of the Su-

preme Court (STF) to a request under Request Category 1. The court’s response stated:

“I inform you that the amounts paid under the per-diem rubric are the

responsibility of the Department of Administration and Finance [Secretaria

de Administração e Finanças].”

In this case, the person responsible for the answer recognized that the per-

diem information was relevant and pertinent and directed the user to the depart-

ment responsible for the information. This practice, despite complying with Article

11, § 1, section III of the FOI law, does not comply with the best practice identified in

the audit – forwarding of the request to the appropriate entitiy1.

A similar situation occurred twice in the states of Rio Grande do Norte (RN)

and Sergipe (SE). For example, the TJ-RN states:

“This General Justice Ombudsman [Ouvidoria Geral de Justiça] has

its purpose described in Res. 026/20102. Thus, as a permanent channel of

1 Article 11. The public body or entity must authorize or provide immediate access to the information available. § 1 If it were not possible to provide immediate access, as described above, the body or entity receiving the request must, within 20 (twenty) days: III – inform the citizen that the body or entity does not possess the infor-mation, indicate, if within its knowledge, the body of organ that has the information, or even submit the request to such body or entity, informing the citizen that the information request has been forwarded.

2 Available at: http://sistemasdis-01.tjrn.jus.br/servidorDeArquivos/pages/ documentos/4112.pdf [Accessed on: 30.10.2014]

105BARRIERS TO TRANSPARENCY AND ACCESS

intercommunication, we inform you, Sir, that your request does not pertain

to this Ombudsman. The transparency portal of the Judicial Branch has the

information requested.”

In this case a legal norm is cited as a justification for not providing the infor-

mation. However, the referenced legal norm, which predates the FOI law, establishes

in its Article 3, § 1 that it behooves the ombudsman to “appropriately engage the

various entities of the judicial branch in the search for information, which will be

analyzed, evaluated and turned over to users whenever requested.”

With the signing of Law 12.527/11, the TJ-RN issued a norm specifically addres-

sing the FOI law: Res. 016/20123. This norm designates the ombudsman as the en-

tity responsible for receiving and handling FOI requests, until the approval of state

legislation addressing the subject. Thus it remains unclear to a citizen searching for

information which bureaucratic procedure to follow. The final result in this case was

that an email was received a few days after making the request:

“In response to case no. 2014-1712 registered with the Ombudsman,

we inform you, Sir, that following numerous attempts to contact you via te-

lephone on the 9th day of this month and this year, we were informed that

the contact number does not belong to you but rather to a Mrs. Juliana.

Therefore, based on Art. 10, III, B, we consider this request closed.”

In this instance, the case was closed despite having other forms of contact

available4. Moreover, it is not possible to deduce to which norm “Art. 10, III, B” refers.

Bureaucratic barriers were also identified with the Courts of Justice (Tribunais

de Justiça) of the states of Parana and Sergipe. The former responded to all the

requests submitted, but only one requests was answered accurately (Request Cate-

gory 3). The other responses stated that the court did not accept requests in digital

format; only in-person or via regular mail, the postal address for which was supplied.

This practice is against the rules of Law 12.527/11, which state:

3 Available at: http://sistemasdis-01.tjrn.jus.br/servidorDeArquivos/pages/ documentos/4112.pdf

4 It should be noted that the provision of various forms of contact was beyond the requirements of Article 10 of the FOI law, as well as of Article 12 of Decree 7.724/2012.



“Art. 10. Any interested party can submit freedom of information re-

quests to the bodies and entities references in Article 1 of this law, by any

legitimate means. Such requests must contain the applicant’s identification

and description of the required information.

[…]

§ 2 The public bodies and entities must enable an alternative to sub-

mit freedom of information requests through official sites on the Internet.”

In relation to the second category of bureaucratic barriers (i.e., that a particular

request was not submitted in an appropriate format or did not comply with a formal

requirement), we encountered some cases in which the rejection of the requests was

based on the supposed absence of required information, such as a justification for

making the request. As previously explained, this requirement is explicitly prohibited

under the FOI law.

Examples of this type of case include the three rejections by the TJ-SC, and

one rejection by the TJ-MT (regarding Request Category 4). The latter provided the

following response:

“Regarding the requested minutes, the interested party in waiting

must provide the relevant bodies of this court with the explanatory reasons

that justify the request.”

2. Legal and regulatory barriers, and misuse of exemptions

2.1. Excessively broad interpretation of exemptions

Brazilian legislation provides for limited exceptions from disclosure in whi-

ch freedom of information may be refused or curtailed. These exceptions can be

summarized as follows:


Fiscal or judicial secrecy, as per Article 22 of the FOI law5;

Information sensitive to national security, as per Article 23 of the FOI law6;

Personal information, which might affect the intimacy, privacy, honor or image

of those involved, as provided by Article 31 of the FOI law7.

In addition to the cases included in Law 12.527/11, Decree No. 7724/2012 (which

regulates the FOI law within the federal executive branch) includes three types of

requests which may be rejected. These include requests that are: i) generic, ii) dis-

proportionate or unreasonable, iii) require additional work, analysis, interpretation

or consolidation of data and information, or require the production or processing of

data outside the competency of the agency or entity.

As the following pages show, a potentially overbroad interpretation of these

exceptions has been used to justify the refusal of legitimate FOI requests.

2.1.1. Disproportionate or unreasonable requests

Within this rubric, the aforementioned case of the TJ-SC is notable. Of seven

requests submitted, the TJ-SC rejected three, and did not respond to two. In all

three rejections the court offered the same justification and narrative, changing only

specific issues for each request and the date of the decision. The argument that a

request was disproportionate, generic, or that it demanded additional work was

frequently presented in notices of rejection: for example, in the rejections by the

5 Art. 22. The provisions of this law do not exclude the other legal hypotheses of legal confidentiality and secrecy and judicial secrecy, nor the hypothesis of trade secret arising from the direct use of economic activity by the state or by individuals or private entities that have any link with the government.

6 Art. 23. The security of society and the state are considered essential, therefore it issubject to secret classification, information for which its dissemination or access could: I – endanger the de-fense and national sovereignty or the integrity of the national territory; II – harm or jeopardize the conduct of negotiations or international relations of the country, or that have been provided in confidence by other states and international organizations; III – endanger the life, security, or health of the population; IV – present high risk to the financial, economic or monetary stability of the country; V– harm or pose a risk to strategic plans or operations of the armed forces; VI – harm or pose a risk to research projects and scientific and technological development, as well as systems, goods, facilities or areas of national strategic interest; VII – endanger the safety of institutions and domestic or foreign authorities and their families; or VIII – compromise intelligence ac-tivities, as well as investigation or surveillance in progress, related to the prevention or prosecution of offenses.

7 Article 19, Balanço de 1 ano da Lei de Acesso à Informação Pública: acesso à informação e os órgãos de justiça brasileiros – 2012-2013. São Paulo: Article 19, 2013. Available at: http://artigo19.org/wp-content/uplo-ads/2014/01/balanco_1_ano_lei_ acesso.pdf [Accessed on: 03.11.2014]



TJ-DFT, and Regional Labor Court (TRT) Region 1, to Request Category 2. Another

notable response of rejection on the grounds of disproportionate work was the TRF

Region 4’s argument that staff limitations made it impossible to meet the reques-

ter’s “private” research needs:

“As a matter of policy, public servants of this Court are not available to

conduct private research. The human capacity of this public body, except

for performance of the necessary administrative activities, is primarily en-

gaged in providing support for the main activity of any court, the delivery

of judicial services.” (Request category 1)8

It hardly bears asserting that public bodies have a responsibility to meet the de-

mands of passive transparency as stipulated by law. Accordingly, ensuring complian-

ce with this should be incorporated into the administrative routines of the judiciary.

It is also important to highlight the fact that two courts (the TST and TJ-RR)

answered the same question accurately and in a period considered reasonable. The

example of these courts seems to indicate that in principle such requests can be

answered without placing great burdens on entities.

2.1.2. Generic requests or requests requiring additional work

In the case of negative responses in Request Category 4, the TRF Region 2 and

TJ-SC both argued that the requests were generic and disproportionate. The TRF

Region 2 states:

“As can be easily seen, the requests for information (a) and (b) the

email included below are generic. Therefore, based on DECREE No. 7.724

OF MAY 16, 2012, which regulates the aforementioned statute, particularly

article 12, III and article 12, it is legally impossible to fulfill the request.” (em-

phasis in the original)

Meanwhile, the response of the TJ-SC reads as follows:

8 Despite providing an inadequate response to this request, the TRF Region 4 provided accurate responses to four of the seven requests submitted.


“In this case, compiling the requested information would be burden-

some, given the extensive list of names of the judges that were promoted.

It should also be pointed out that the interested party did not even indicate

the period requested.”

The TJ-SC’s complaint regarding the volume of information requested was also

highlighted by the TJ-RR in its initial response. On appeal, however, the latter rever-

sed its decision, providing the requested information. Cases such as these demons-

trate the broad spectrum of interpretations employed by public servants, which ren-

der access to information more discretionary than it should be.

2.2. Misinterpretation of exceptions; active transparency and the format of documents

As previously discussed, Article 11 of the FOI law stipulates that in cases where

the information is already available to the public, the entity must indicate precise

steps for the user to follow in order to obtain the information. This is not a valid

exception for rejecting a request, but rather an instruction on how to provide the

information in such cases9.

Some requests in this audit asked for information already public, but requested

in a specific format (i.e., Requests 1 and 7, which asked for files processable by com-

puter) or in a different aggregation to that originally published (Request Category

2). All these requests concerned the remuneration of judges10.

Some of the requests were denied on the basis of justifications presented in

the previous section; however, others were answered inaccurately in that they simply

made reference to the transparency portals and ignored the specificity of a request.

9 As discussed below, such directions are not always observed by the courts, which, when indicating that the information is available on the transparency portals, fail to instruct the user on how to obtain it.

10 In order to improve the formulation of the requests in this audit, we conducted an overview of active transparency practices about the information available on the remuneration of judges and the courts’ servers. The practices are very heterogeneous, with different ways of presenting data, and of aggregation and format possibilities. It should be noted that only the TJ-RJ, TJ-MG, TJ-BA provide data processable by computer on their websites.



2.2.1. Requests for information already available in active transparency platforms

In some cases, courts have held that compliance with the rules of active trans-

parency exempted them from having to provide the same information in their res-

ponses. One example can be found in the response of the TJ-RS, which at first direc-

ted the user to the transparency portal where it is possible to do a search per month.

In response to an appeal, the court stated:

“We reiterate the points of the previous communication, stating that

we do not have the data in the format requested, and noting that the TJRS

uses a similar tool [modelo de disponibilização] to the one used by the

National Judicial Council (CNJ). Moreover, the data available in the trans-

parency site follows the recommendations of the National Judicial Coun-

cil’s (CNJ) Resolution No. 151, and is collected monthly, directly from the

payment stubs of public servants and judges, with no spreadsheet in the

requested format, which can be assembled by interested parties from the

information published on the transparency site”.

It is important to note that the transparency tool mentioned in the response

presents the data in HTML format, as determined by the National Judicial Council

(CNJ). However, this format makes the analysis of data difficult; one of the reasons

we issued the requests in the first place. Moreover, the tool does not make data avai-

lable for 2012, the time period specifically requested11.

In response to the same question, the TJ-GO stated “all the requested data is

freely accessible to the general public through the Transparency Portal, the world

wide web (Internet)”, offering no additional information on how to access said data.

A similar response was received from the TJ-DFT, which said, in response to

the same question: “It is our understanding, unless otherwise proven, that there is

no need for this Sub-secretary to provide the information requested, since the data

is already available on the Transparency Portal”.

11 The request stated: “I hereby would like to request, based on Law 12.527, the following information: Con-sistent with the provisions of the only Annex of Resolution No. 151 of the National Judicial Council (Conselho Nacional de Justiça – CNJ) I request that the Court provide a table in a computer-processable format, following the pattern of the resolution of the National Judicial Council (CNJ), summarized on an annual basis for the years 2012 and 2013, the amounts paid to each judge in the form of base salary, personal benefits, subsidies, allowa-nces and other discretionary benefits. The table should provide the total annual amount of the compensation elements paid to each judge of this Court during the years 2012 and 2013.”


Again, there are some problems with this response. Firstly, it did not clearly

indicate the precise location where citizens could obtain the data requested, only

the generic address of the portal. Secondly, the availability of information only in a

format different to that requested was used to exempt the body from responsibility

for responding to the request. The result is that citizens do not get the information

in the format and aggregation pattern required.

The TRT Region 15 provides the webpage address, or URL (Uniform Resource

Locator), to the site containing the information requested: however, this URL pro-

vides more than 200 pages, and PDFs with monthly figures. In regards to this data,

the court states:

“We emphasize that the values provided by TRT 15 in its ‘Transpa-

rency’ pages are in accordance with the National Judicial Council (CNJ)

Resolution No. 102/2009, as amended by National Judicial Council (CNJ)

Resolution No. 151/2012.”

Given these examples, it is important to emphasize that the issue of format is no

minor detail. It is the format which often determines a citizen’s ability to obtain the

requested data and conduct cross-tabulations and comparisons to give meaning to

the raw and disaggregated values. This was precisely the point of Request Category 1.

For the second request, which asked for a publicly unavailable breakdown of

data, responses were often similar. In addition to not providing the correct format,

responses did not address the most basic questions of applicants. The TRT Region

10, for example, in its initial response stated:

“[...] I inform you that the data requested by the interested party

should be accessed on the Court’s transparency page, where they are pu-

blished in the format determined by Resolution No. 102 of the National

Judicial Council, as amended by Resolution No. 151 of the same entity.”

Upon appeal, the court responded by reiterating its interpretation of complian-

ce with active transparency standards, insisting it was the responsibility of the citi-

zen to organize the data. In the words of the court:



“The technical body affirms, categorically, that it is in compliance with

the stipulations of Res. 102 of the National Judicial Council (CNJ), as amen-

ded by Res. 151 of the same entity, norms which are consistent with Law

12.527/11. The decision made does not merit mending due, mostly, to its full

regulatory compliance. The interpretation of the requested data is the res-

ponsibility of the applicant, independently of the format in which the data

is presented, provided that it meets the legal norms, as characterized here.”

What can be observed in this type of response is that the transparency portals

can serve as justification for the denial of FOI requests. The prevailing interpretation

of the courts is worrying, since it seems to infer that their obligations regarding FOI

requests are limited to complying with the standards of the National Judicial Council

(CNJ), ignoring the possibility that citizens may request additional data or data in

other formats.

2.2.2. Questions regarding data format

This situation is evident in Request Category 7, in which the applicant explicitly

specified being aware of the existing active transparency requirements and justified

the request based on the FOI law:

“I am aware that in the Transparency Portal of this court, the details

of the payroll of judges by month are already available in the format stipu-

lated by Resolution 102 of the National Judicial Council.

In view of this, and based on Law 12.527, I would like to request spre-

adsheets detailing the pay of judges for each month during the years 2012

and 2013, in a format processable by computer, which can be .xls or .xlsx

(excel), or some other format processable by computer. After all, the for-

mat used by the Transparency Portal of this court, that is, .pdf, is not com-

puter-processable, preventing proper data analysis. […]

In this case, the response rate was significantly lower than for the other ques-

tion categories, and reasons given for not supplying information were varied: the

TJ-BA, for example, stated that:


“To obtain information in files with the specific formats requested it is

necessary to directly contact the responsible department, in this case the

Human Resources Department, by telephone: (71) 3372-1649 / 3372-1665 /

3372-1650, to find a better solution to your problem.”

The question is raised here as to why the department responsible for proces-

sing FOI requests did not forward this one to the correct department, especially

since the response implies knowledge of the information requested and its location.

The TJ-MG, in turn, stated that it uses the PDF format, which could be exported

into a textediting format. While this assertion may be true, the files are presented on

a monthly basis and contain large numbers of pages, rendering information extre-

mely difficult to process. The court justified its decision as follows:

“The adoption, by the Tribunal de Justiça do Estado de Minas Gerais

[Court of Justice of the State of Minas Gerais], of documents in PDF format

for the publication of information regarding payroll, is based on the need

to preserve the credibility and reliability of such data.”

This response seems contradictory: on one hand it states the PDF format en-

sures the data’s reliability while, on the other, it states that this format allows for the

export of data into editable text. If the latter is true, what would then prevent a user

from exporting the data into text, and creating a new PDF identical to the original,

but with adulterated data? The issue of credibility and reliability is crucial and must

be addressed in the dialog and norms regulating freedom of information and trans-

parency. However in this case, it does not justify a rejection of the request. There are

mechanisms available such as electronic signatures, for various formats, which could

help address the issues of credibility and reliability.

The court’s interpretation is that it is complying with the stipulations of the FOI

law, as can be read in its response:

“The published PDF documents comply with the norms established by

the Freedom of Information Law, item II of § 3 of Art. 8, which states ‘such as

spreadsheets and text’, and presents a non-exhaustive list of suggestions.”



Again, this response is of little relevance to the average user, and appears to

serve only as justification for failure to provide the spreadsheet requested.

It should be noted that the format of remuneration data presented in the

court’s active transparency portal is fragmented and does not allow for a historical

and comparative analysis. Search tools and filters, for example, at times restrict que-

ries to one month or judge per search.

In general, while the aforementioned FOI requests were not explicitly denied, the

courts only provided data already available in the active transparency portals, ignoring

the specificity of the applicant’s request for a different format or level of aggregation.

This is particularly true for responses to Requests 1, 2 and 7. Some examples follow:

In the case of Request Category 1, the TRT Region 15 directed the user to its

transparency portal, where, although it was possible to obtain monthly files

in PDF format, each file had an average of 100 pages and did not include the

judges’ names (only giving compensation figures by position). In this particular

case, not only was the information incomplete, but its presentation precluded

any processing of the data which might permit a more in-depth analysis.

The TJ-MT complied with part of the request by describing the types of

allowances, personal benefits, and discretionary benefits received by judges.

However, the court directed the user to its website, where data is available

by month in an HTML format. The website did not permit downloading of the

data aggregated (either by month or year) for further computer processing.

Furthermore, Resolution 151/2012 only permits the viewing of data pertaining

to one judge; to obtain an aggregate overview of all the judges it would be

necessary to download information, one by one, for each of the 250-plus judges

of the court.

This practice is observed in numerous courts including the TJ-RS, TJ-RO, and

TRF Region 2, among others. Again and again the courts’ responses claim that the

data requested is already available on their respective websites. However, again and

again, the data available on these sites is, as the study shows, fragmented or not

processable by computer.

Request Category 7 was formulated in order to address the issue of receiving

a high number of responses to Requests 1 and 2, which simply directed the users to


the active transparency websites. In Request 7, the interested party explicitly stated

its knowledge about the information already available on the active transparency

sites, and requested data in very specific format (spreadsheets), and level of aggre-

gation. Some courts, such as the Supreme Court (STF), responded to this question

with the justification that their system did not allow for the export of spreadsheets.

Other courts, such as TJ-MG and TJ-BA, justified their response on the alleged im-

possibility of submitting the data in a format processable by computer based on

Article 4 of Law 12.527/11.

Another notable response to the requests for data in specific formats came

from the TJ-ES, which claimed its format, “is not only processable but also editable”;

moreover, that authorization would be required “from a superior”, before a response

could be given.

The response by the TJ-CE merits special mention. Regarding the request for

norms and minutes addressing the promotion of judges, the TJ-CE recommended

the purchase of a book “Division Code and Judiciary Organization of the Judicial

Branch of the State of Ceará” (Código de Divisão e Organização Judiciária do Poder

Judiciário Estadual Cearense) available in bookstores. Alternatively one might con-

sult the book on the court’s website: the response did not contain a link to the file,

or to the specific information requested.



chApter 7evAluAting the plAtforms used to mAnAge freedom of informAtion requests

The evaluation of platforms involved identifying and assessing the tools pro-

vided to citizens, by which they can submit FOI requests to the courts via Internet.

This audit used the same criteria as the General Audit, namely: (i) the existence of a

platform dedicated specifically to the submission of FOI requests; (ii) confirmation

of receipt of the request and notification when the request is answered; (iii) the pos-

sibility of tracking the request; and (iv) a section for filing appeals.

Of the 40 courts analyzed, 33 of them (82.5%) were found to have no formal platform

to handle FOI requests; instead, this responsibility was transferred to the ombudsman.

In many cases, the ombudsman’s site makes available a list of options, such as

‘Complaints’, ‘Suggestions’, and ‘Talk to the Inspector’, but does not display an option

for ‘Access to Information’. We believe employing the ombudsman’s website for FOI

requests can generate considerable problems, which we outline in the following pages.

Law 12.527/11 stipulates specific rules in the procedures for requesting informa-

tion; procedures which are not observed by the ombudsman websites. An example

is the right to appeal decisions rejecting a FOI request. There is no specific tool for

filing appeals and the citizen has no alternative but to submit a new request. This

complicates the process and may confuse the user, who might then simply give up,

due to the lack of appropriate fields.

Many ombudsman’s sites have character limits, which can become barriers to

passive transparency. This is the case of the TJ-BA, TJ-RS, TJ-TO, and the Superior

Labor Court (Tribunal Superior do Trabalho), the TST.

Some platforms require the user to provide a telephone number. While this

might conceivably facilitate contact with a citizen, it might also represent an undue

requirement of the applicant, since Law 12.527/11 asks only for simple identification;

name and CPF number (similar to a social security number), or ID (Registro Geral or

RG). As we saw earlier in the case of the TJ-RN, the court rejected the request based

on the grounds that the telephone number provided was incorrect.



The practice of the TJ-SE should also be noted, where serious navigation dif-

ficulties were encountered. The TJ-SE’s website features a link to the ombudsman’s

site, but there is no easily identifiable link on the main page for a form concerning

FOI requests. Requests sent to the ombudsman, however, are returned with the

following message:

“The Ombudsman is not the appropriate entity to provide clarifica-

tions related to the Freedom of Information Law for this Court of Justice.

Based on the attached Articles 3 and 5 of the Ordinance 32/2012 of the

JT/SE such duties are the responsibility of the Citizen Services Division

[Divisão de Serviços ao Cidadão]. This time, please submit your request

through the appropriate channel, which can be found on the home page of

the Court of Justice of Sergipe [Tribunal de Justiça de Sergipe] available

at: Menu Principal>Transparência>Solicitação de Informação or directly at:

http://www.tjse.jus.br/tjnet/solicitaInformacao/solicitacao.wsp”

Despite the accuracy of the response sent by TJ-SE regarding the precise loca-

tion of the form to submit FOI requests, the link is difficult to access and the process

complex: all citizens currently have to enter this process with the ombudsman in

order to find the correct procedure for submitting requests.

Another issue in relation to the TJ-SE platform, was the malfunctioning of the

FOI request platform, which sometimes returned a ‘Not Found’ message: the same

problem occurred with the TJ-DFT.

Other sites also had technical problems when FOI requests where submitted.

These include the TJ-RS, TJ-AC, TJ-RO, TJ-AM, and the STJ.

The General Audit and the Judicial Branch Audit conducted by the Public

Transparency Program, coordinated by FGV EBAPE and CTS at FGV DIREITO RIO,

illustrate considerable heterogeneity in commitments to transparency.

The results of both audits suggest the need for new and renewed efforts to

fully comply with the legal obligation of Brazil’s new freedom of information (FOI)

law – 12.527/11. Response rates for the General Audit and the Judicial Branch Audit

were fairly similar: 69% and 61%, respectively. These figures show that, on average,

one out of three requests was completely ignored. The accuracy rates of responses

– a more important metric because it measured the percentage of directly relevant

answers – were significantly lower: 57% for the General Audit, and an alarming 26%

for the Judicial Branch Audit.

However, although these numbers are relatively low, the commitment to trans-

parency by Brazilian public entities is not monolithically negative. Two and a half

years after the FOI law came into effect, significant differences in compliance are vi-

sible across the public sector. These range from moderately good adherence to the

law by the federal government and the State and City of São Paulo to extraordinarily

poor compliance by the State and City of Rio de Janeiro.

The Judicial Branch Audit clearly showed how institutional commitments are

crucial for implementation and compliance. For example, among the courts with the

best results is the State Tribunal of Roraima; unexpected of a small rural state with

a budget a fraction of the size of courts in wealthier regions. In this sense, countries

such as India and Mexico have shown the world that, sometimes, guaranteeing the

right of access to information has little to do with wealth or bureaucratic sophistica-

tion, and more to do with leadership and commitment.

conclusion



The Scope of Research

Combined, both audits included 717 FOI requests: 453 in the General Audit and

264 in the Judicial Branch Audit. These requests were submitted to 173 different

Brazilian public entities.

In the General Audit, requests were submitted based on 55 questions to 133

public bodies belonging to all the branches of government in eight jurisdictions:

the states of São Paulo, Minas Gerais and Rio de Janeiro, as well as their respective

capitals, in addition to the Federal District and the federal government. We chose to

evaluate a larger number of entities in different branches of government on a wide

range of topics. The General Audit, in this sense, is exploratory in nature and ser-

ves to provide an overview of salient positive and negative trends, including overall

commitments to public transparency and to Law 12.527/11.

We submitted virtually the same seven requests to 40 courts of justice as part

of the Judicial Branch Audit. The courts included three Superior Courts (Tribunais Su-

periores), the Supreme Court (Supremo Tribunal Federal, or STF), the Superior Court

of Justice (Supremo Tribunal Federal or STJ) and the Superior Labor Court (Tribunal

Superior do Trabalho (TST), five Federal Regional Courts (Tribunais Regionais Fede-

rais), five Regional Labor Courts (Tribunais Regionais do Trabalho), 26 State Courts

(Tribunais de Justiça dos Estados), in addition to the Court of Justice of the Federal

District and Territories (Tribunal de Justiça do Distrito Federal e Territórios).

Our request topics were selected to reflect current challenges to Brazil and

the Brazilian courts. These topics included the remuneration and promotion of jud-

ges, the prevention of nepotism, and the administrative management of courts. The

information obtained in the course of this research serves as a first step for future

work regarding the judiciary.

Methodology

We designed the methodology to be academically rigorous, but also straigh-

tforward and replicable. Replicability is essential so that the findings of transparency

audits conducted by national and international entities can be compared with each

other. Comparison of findings across audits is currently not possible due to the di-

versity of definitions and metrics used.

Within this context, the Transparency Audit Network is attempting to build a

121CONCLUSION

network of national and international institutions interested in public transparency,

which can conduct parallel audits using the same metrics, to better enable the com-

parison of results across institutions. The Transparency Audit Network is an initiative

supported by the Open Society Foundation (OSF) and coordinated by the Public

Transparency Program, in turn coordinated by EBAPE and CTS at FGV DIREITO RIO.

As detailed in Chapter 1, we selected three straightforward metrics for the analysis:

response rate, accuracy rate, and average response time. We also included a field

experiment in each audit. The first field experiment in the General Audit tested whe-

ther discriminatory practices existed in the provision of access to public information.

The second field experiment in the Judicial Branch Audit sought to identify whether

justified FOI requests (i.e., requests which explained the reasons for the request)

would receive more satisfactory responses than requests with no justification.

The field experiments generated intriguing results, although further tests are

needed to confirm the inferences derived. The most remarkable results were obser-

ved when analyzing two sets of male and female identities. If one is to perform an

Internet search of the first set, one will find individuals with clear institutional affilia-

tions to the FGV (the ‘institutional identities’). If the same procedure is followed for

the second set of identities, no identifiable profile on the World Wide Web is found

(the ‘non-institutional identities’). These two sets of users submitted 322 requests

and received considerably different responses. The two users with non-institutional

identities received approximately 10% fewer responses than those with institutional

identities a statistically significant result. Non-institutional identities also received

8% fewer accurate responses. In addition, the researchers of the FGV received their

responses in 17.5 days on average, compared to the pair with no clear institutional

affiliations, who received responses eight days later on average, in 25.5 days.

These findings reinforce the hypothesis that there may be discriminatory practi-

ces in the processing of FOI requests based on the institutional affiliation of the inte-

rested party. They provide good reason to reconsider the legal requirement that inte-

rested parties must reveal their identity, as stipulated under Article 10 of Law 12.527/11.

Such legal requirements, which can be encountered around the world, may be doing

more harm than good when it comes to maximizing public information rights.

The digital platforms and communication methods for submitting FOI requests,

responses, and appeals were also evaluated. The platform e-SIC, developed by the

Comptroller General of the Union (Controladoria Geral da União, or CGU) deserves



special commendation. This platform has numerous functions that facilitate the FOI

requesting process.

It should be noted that the international community is moving toward the use

of similar platforms. Mexico, for example, will soon employ a single portal: for making

FOI requests for appealing and for responding to FOI petitions in all branches and

levels of government. Considering that Brazil’s e-SIC platform is freely available to

any public entity, there can be few justifications for using ombudsmen’s platforms

or the ‘Contact Us’ pages, both of which are inadequate for submitting and handling

FOI requests. The lack of dedicated platforms for FOI is particularly troublesome in

the case of the judiciary since only a very small number of courts have specific pas-

sive transparency platforms.

Institutional challenges

This analysis identified several gaps with regards to the fulfillment of the right

of freedom of information. Section II of Law 12.527/11 stipulates that on receiving

an unsatisfactory response, requesters can submit up to two appeals. However, the

filing of a third, to a dedicated entity for handling such appeals, is only possible at

the federal level, in which the Comptroller General (CGU) is the relevant body. Under

the current framework, third-level appeals to independent entities are not possible

at the federal, state, or municipal level within the judicial and legislative branches.

Other institutional challenges identified in the analysis, relate to the way in which

the various jurisdictions regulate Law 12.527/11. Many states as yet do not regulate the

implementation of the FOI law. On the other hand, some states, such as Rio de Janei-

ro, have introduced illegal regulations. Here, the regulations preclude the submission

of requests via digital means and also require petitioners to sign an absurd statement

of liability. These requirements create serious limitations to exercising the right for

freedom of information, and can only be interpreted as purposively antagonistic.

On a final note, the identification requirement, as previously mentioned, may

represent a serious disincentive for practicing freedom of information rights. It is not

difficult to imagine applicants being intimidated or coerced in small municipalities,

where they can easily be identified.

Given the above challenges, we consider it necessary to strengthen the FOI

law through concerted efforts. Institutions such as the Public Prosecutor (Ministério

Público) can and should expand their efforts to enforce the Law 12.527/11.

123CONCLUSION

General recommendations

Throughout this report, the Program for Public Transparency, coordinated by

EBAPE and the CTS FGV RIGHT RIO, identified a number of best practices and

international ‘gold-standards’ that can serve as a model for Brazilian government

agencies. We also identified obstacles to public transparency that must be rectified.

Below we present some of our principal recommendations in summary form:

Revoke or modify state and local ordinances, which establish procedures limiting

the fundamental right of freedom of information, such as the requirement that

the request be made or the answer sought, in person.

Revoke or modify rules requiring the applicant to provide personal identification,

as this creates the potential for discrimination and intimidation. In an ideal

scenario, the citizen should be required to inform only a means of contact, as

indicated by the Model Inter-American Law on Access to Information of the

Organization of American States.

The Institute for Access to Public Information and Data Protection (Instituto

Federal de Acceso a la Información y Protección de Datos, or IFAI), in Mexico

provides a prime example for Brazil to follow with regards to oversight and,

institutionalization of the FOI law. Unlike Mexico, Brazil lacks a regulatory

agency with the power to regulate and enforce the FOI law in all spheres of

government (i.e., at federal, state and municipal level, as well as all branches and

autonomous bodies).

Create new or improved channels for requesting information by citizens,

observing the provisions of FOI law. Of note here is the need to avoid

unnecessary or excessive bureaucratic obstacles that may prevent or limit the

exercise of the right of freedom of information. Preference should be given

to the platform e-SIC, in place of ombudsman services or ‘Contact Us’ pages,

which insufficiently address obligations imposed by the Law 12.527/11.

The various bodies and authorities should take steps to ensure that legitimate

claims are not rejected based on broad and unreasonable interpretation of the

‘additional work’ clause.

Delimit the range of possible interpretations of exceptions to the right of

freedom of information.



Government should respond to public information requests at the level of detail

requested by citizens. There should be compliance with the FOI law, with regard

to the use of open and non-proprietary formats, and adoption of the principles

of open data across the board, including passive transparency, according to the

recommendations of international bodies and the commitment made by Brazil

to the Open Government Partnership.

Specific recommendations for the Judicial Branch

As we saw earlier, Resolution No. 102 of the National Judicial Council (CNJ),

addresses the issues of active transparency practices, with regard to budgetary and,

financial data, and personnel and remuneration structures. This norm is important,

and a positive step towards detailing the transparency obligations of the entities of

the judiciary. The results of the judicial evaluation show however, that this resolution

is often interpreted as a justification for not responding to FOI requests targeting

information not included in the resolution. Moreover, the resolution is also used as

a justification for rejecting requests asking for data in formats and/or aggregation

levels different to those established by the resolution. Both of these practices are

against the FOI law.

Considering that one goal of Resolution 102 is to “standardize the presenta-

tion of information in order to make it intelligible and comparable for the benefit of

effective public oversight of public spending”, the following changes are proposed:

A reformulation of Article 4, paragraph 3, of Resolution 102 of the National

Judicial Council (CNJ). This article states that information must be published

in hypertext (HTML) format, to fit the provisions of Article 8, § 3, sections II

and III of the FOI law. The aforementioned sections of the FOI law enable “the

recording of reports in various electronic formats, including open and non-

proprietary, such as spreadsheets and text, in order to facilitate the analysis of

information”, and “the automated access by external systems in open formats,

structured and readable by computer.”

A modifying of the resolution to permit data queries relating to remuneration

not only on an individual, and month-to-month basis, but also on remuneration

aggregated annually and by employee. The data should be made available in a

way that permits its processing and analysis.

125CONCLUSION

It is important to note that the aforementioned resolution is clear in its purpose

and that the National Judicial Council (CNJ) could not anticipate its interpretation

by the courts. However, evidence presented herein makes clear the limitations in

implementation. It is therefore essential to review the points listed above in order to

align judicial bodies with best practices of promoting and complying to freedom of

information provision. It is also important that the National Judicial Council (CNJ)

monitor and enforce the active transparency practices of the judiciary for which, as

we have seen, there is significant heterogeneity.

Next steps

Law 12.527/11 represents a significant achievement in the legislative history of

the country. However, the task of changing the culture of secrecy that prevailed

for decades will not be instantaneous. It is not enough to introduce a Freedom of

Information Law, for public transparency to become a reality. There must also be

both political and administrative commitments, and the broader use of the law by

civil society.

This study should suggest that grounds for conducting further research and

evaluation on transparency are very fertile. This study adhered to one of many pos-

sible research designs, and there is a need for more research to be undertaken in

order to understand the workings of FOI law in other regions, especially outside

the major metropolitan centers. Likewise, ample space exists for conducting spe-

cific assessments by sector, in areas such as health, education, public safety, and

urban mobility, among others. Finally, the transparency of some agencies has not

been scrutinized in sufficient detail. Such entities include state and local prosecu-

tors, entities of the indirect public administration, and also private and non-profit

entities receiving state funds. The Program for Public Transparency at the FGV aims

to undertake concerted efforts towards a better understanding of compliance with

Brazil’s new FOI law. In doing so, it hopes to contribute to greater commitments to

transparency in Brazil and internationally.



ALIANZA REGIONAL POR LA LIBRE EXPRESIÓN E INFORMACIÓN. Saber Más

VI: Informe Regional Sobre Acceso a Información Pública y Apelaciones. 2014.

Available at: http://www.alianzaregional.net/ acciones/investigacion-aplicada/saber-

masvi/

ARTIGO 19. Balanço de 1 Ano da Lei de Acesso à Informação Pública. São

Paulo: Artigo 19, 2013.

BEGHIN, N. & ZIGONI, C. (orgs), Avaliando os websites de transparência orça-

mentária nacionais e sub-nacionais e medindo impactos de dados abertos sobre

direitos humanos no Brasil. Brasília: Instituto de Estudos Socioeconômicos, 2014. Avai-

lable at: http://www.inesc. org.br/biblioteca/publicacoes/textos/pesquisa-transparen-

cia-orcamentaria-nos-websites-nacionais-e-sub-nacionais [Accessed on: 29.10.2014]

BRASIL. Lei 12.527 de 18 de novembro de 2011. Available at: http:// www.pla-

nalto.gov.br/ccivil_03/_ato2011-2014/2011/lei/l12527.htm

BRASIL. Resolução ANTT nº 442, de 2004. Available at: http://appweb2.antt.

gov.br/resolucoes/00500/resolucao442_2004.htm

BRASÍLIA. Decreto Federal 7.724. Available at: http://www.planalto. gov.br/

ccivil_03/_ato2011-2014/2012/decreto/D7724.htm

CONSELHO SUPERIOR DA JUSTIÇA DO TRABALHO (Brasil). Resolução n. 107,

de 29 de junho de 2012. Diário Eletrônico da Justiça do Trabalho, Brasília, DF, n. 1011, 2

jul. 2012. Caderno Jurídico do Conselho Superior da Justiça do Trabalho, p. 1-3. Availa-

ble at: http:// aplicacao.tst.jus.br/dspace/handle/1939/24937 [Accessed on: 03.11.2014]

ESTADO DE MINAS GERAIS. Deliberação 2.555/2013. Available at: http://

www.jusbrasil.com.br/diarios/54195134/al-mg-11-05-2013-pg-13

references



ESTADO DO RIO DE JANEIRO. Decreto Estadual 43.597 de 16 de maio de

2012. Available at: http://www.aperj.rj.gov.br/legislacao/ dec43597.htm

ESTADO DO RIO DE JANEIRO. Resolução nº 275 de 2013 do Tribunal de Con-

tas do Estado do Rio de Janeiro. Available at: http://www. mpmg.mp.br/acesso-a-

-informacao/requerimento-de-informacao/

FALCÃO, J., HARTMANN, I. A. e CHAVES, V. III Relatório Supremo em Núme-

ros: o Supremo e o Tempo. Rio de Janeiro: Escola de Direito do Rio de Janeiro da

Fundação Getulio Vargas, 2014.

KNILL, C. Introduction: Cross-national policy convergence: concepts, approaches

and explanatory factors. Journal of European Public Policy, v. 12, n. 5, p. 764–774,

2005.

LABORATÓRIO BRASILEIRO DE CULTURA DIGITAL & NICBr, Manual dos Da-

dos Abertos: Governo. Traduzido e adaptado do Open Data Manual – W3C (Co-

laboração: Comunidade Transparência Hacker e Everton Zanella Alvarenga). São

Paulo: Comitê Gestor da INternet no Brasil, 2011. Available at :http://www.w3c.br/

pub/Materiais/ PublicacoesW3C/Manual_Dados_Abertos_WEB.pdf [Accesed on:

02.11.2014].

LANDIS, J. R., AND G. G. KOCH. The Measurement of Observer Agreement for

Categorical Data. Biometrics 33: 159–174, 1977.

MENDEL, T. “The Public’s Right to Know: Principles on Freedom of Informa-

tion Legislation”. London: Article 19, 1999. Available at: http://www.article19.org/

data/files/medialibrary/1797/public-right-to-know.pdf [Accessed on: 03.11.2014].

MINISTÉRIO DA JUSTIÇA, Desafios da Transparência no Sistema de Justiça

Brasileiro. Brasília: Secretaria de Reforma do Judiciário, 2013.

MUNICÍPIO DO RIO DE JANEIRO. Decreto Municipal 35.606 de 2012. Availab-

le at: http://www.rio.rj.gov.br/dlstatic/10112/4179996/4100619/decreto_n_35606.pdf

OAS. Model Inter-American Law on Access to Information. Chapter III (24),

p.12. Available at: http://www.oas.org/dil/access_to_information_model_law.htm.

129REFERENCES

RTI Rating. Countries. Available at: http://www.rti-rating.org/. [Accessed on:

23 de março 2014.]

VIERA, A. J., & GARRETT, J. M. Understanding Interobserver Agreement: The

Kappa Statistic. Fam Med, 37(5), 360-363, 2005.

WEYLAND, K. G. Learning from Foreign Models of Policy Reform. Baltimore:

John Hopkins University Press, 2004.

This book was produced by FGV’s Rio de Janeiro Law School,composed with the Gotham typeface

and printed on offset paper, in 2016.